DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 10/09/2024 is being considered by the examiner.
Specification Objections
The specification is objected to because of the following informalities:
On Page 6, line 2-3, “it is an object of the present invention to focuses on the above-mentioned” should read “it is an object of the present invention to focus on the above-mentioned” in order to avoid a typographical and grammatical issue.
Appropriate correction is required.
Claim Objections
Claims 1-3, 5-16, 18, and 20 are objected to because of the following informalities:
In claim 1, line 1, the term “object detection in a scene; the system” should be changed to “object detection in a scene, the system” in order to avoid a typographical and grammatical issue.
In claim 1, line 3, the term “a memory-suppress diffusion network module adapted to reconstruct” should be changed to “a memory-suppress diffusion network module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 1, line 4, the term “an encoded image; the encoded image” should be changed to “an encoded image, the encoded image” in order to avoid a typographical and grammatical issue.
In claim 1, line 6, the term “a contrastive dissimilarity network adapted to combine” should be changed to “a contrastive dissimilarity network in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 2, line 3, the term “a noise encoding module adapted to generate” should be changed to “a noise encoding module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 2, line 5, the term “a normality memorizing module adapted to integrate” should be changed to “a normality memorizing module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 2, line 6, the term “representations of normality; the set of code memories” should be changed to “representations of normality, the set of code memories” in order to avoid a typographical and grammatical issue.
In claim 2, line 8, the term “a denoise memory-suppress sampling module adapted to reconstruct” should be changed to “a denoise memory-suppress sampling module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 3, line 1-2, the term “with a steadily increasing noise level” should be changed to “with levels” in order to avoid an antecedent issue in succeeding claims.
In claim 5, line 1-2, the term “the noise encoding module is further adapted to sample” should be changed to “the noise encoding module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 6, line 1-2, the term “the normality memorizing module is adapted to transform” should be changed to “the normality memorizing module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 7, line 1-2, the term “the normality memorizing module is further adapted to update” should be changed to “the normality memorizing module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 8, line 1-2, the term “function is used to obtains weights” should be changed to “function is used to obtain weights” in order to avoid a typographical and grammatical issue.
In claim 9, line 1-2, the term “the normality memorizing module is adapted to update” should be changed to “the normality memorizing module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 10, line 1-2, the term “the normality memorizing module is adapted to update” should be changed to “the normality memorizing module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 11, line 1-2, the term “the denoise memory-suppress sampling module is adapted to reconstruct” should be changed to “the denoise memory-suppress sampling module in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 12, line 3, the term “an encoder adapted to encode” should be changed to “an encoder in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 12, line 5, the term “a projector adapted to project” should be changed to “a projector in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 12, line 6, the term “a fusion block adapted to compute” should be changed to “a fusion block in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 13, line 1, the term “system of claim 12 wherein the encoder” should be changed to “system of claim 12, wherein the encoder” in order to avoid a typographical and grammatical issue.
In claim 14, line 2, the term “normalization and ReLU activation” should be changed to “normalization and ReLU (Rectified Linear Unit) activation” as acronyms must be presented with their meanings the first time they are mentioned in the group of claims.
In claim 15, line 1, the term “the system is adapted to provide” should be changed to “the system in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 16, line 1, the term “the system is adapted to generate” should be changed to “the system in order for the subject matter to be positively recited. Please see MPEP 2111.04.
In claim 18, line 1, the term “detecting an foreign object” should be changed to “detecting a foreign object” in order to avoid a typographical and grammatical issue.
In claim 18, line 4, the term “comprising the steps of:” should be changed to “comprising in order to avoid an insufficient antecedent issue and prevent a rejection under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph.
In claim 18, line 8-9, the term “network module and the contrastive dissimilarity network are trained” should be changed to “network module and a contrastive dissimilarity network are trained” in order to avoid an insufficient antecedent issue and prevent a rejection under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph.
In claim 20, line 4, the term “representations of normality; the set of code memories” should be changed to “representations of normality, the set of code memories” in order to avoid a typographical and grammatical issue.
Appropriate correction is required.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action.
This application includes one or more claim limitations that use the word “means” or “step” but are nonetheless not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph because the claim limitation(s) recite(s) sufficient structure, materials, or acts to entirely perform the recited function.
Claims 1, 12, and 18 recite limitations that use words like “means” (or “step”) or similar terms with functional language but do not invoke 35 U.S.C. 112(f):
Claim 1; recites the limitation, “a memory-suppress diffusion network module adapted to…..” [Line 3].
Claim 1; recites the limitation, “a contrastive dissimilarity network adapted to…..” [Line 3].
Claim 12; recites the limitation, “an encoder adapted to…..” [Line 3].
Claim 12; recites the limitation, “a projector adapted to…..” [Line 5].
Claim 18; recites the limitation, “using a memory-suppress diffusion network module…..” [Line 5].
Such claim limitation(s) is/are:
(i) “memory-suppress diffusion network module….” has a structure associated with it a network.
(ii) “contrastive dissimilarity network….” has a structure associated with it a network.
(iii) “encoder….” has a structure associated with it an encoder.
(iv) “projector….” has a structure associated with it a projector.
Because this/these claim limitation(s) is/are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are not being interpreted to cover only the corresponding structure, material, or acts described in the specification as performing the claimed function, and equivalents thereof.
If applicant intends to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to remove the structure, materials, or acts that performs the claimed function; or (2) present a sufficient showing that the claim limitation(s) does/do not recite sufficient structure, materials, or acts to perform the claimed function.
Claims 2, 5-7, and 9-11, recites limitations that use words like “means” (or “step”) or similar terms with functional language and do invoke 35 U.S.C. 112(f):
Claim 2; recites the limitation, “a noise encoding module adapted to…..” [Line 3].
Claim 2; recites the limitation, “a normality memorizing module adapted to…...” [Line 5].
Claim 2; recites the limitation, “a denoise memory-suppress sampling module adapted to……,” [Line 8].
Claim 5; recites the limitation, “the noise encoding module is further adapted to…..” [Line 1-2].
Claim 6; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 7; recites the limitation, “the normality memorizing module is further adapted to…...” [Line 1-2].
Claim 9; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 10; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 11; recites the limitation, “the denoise memory-suppress sampling module is adapted to……,” [Line 1-2].
Claim 11; recites the limitation, “a fusion block adapted to……,” [Line 6].
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
After a careful analysis, as disclosed above, and a careful review of the specification the following limitations in claims 2, 5-7, and 9-11;
(i) “noise encoding module” (Paragraph [0013, 0016, and 0085]- In some embodiments, the memory-suppress diffusion network module further includes a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image, a normality memorizing module adapted to integrate a set of code memories to establish consistent representations of normality, and a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques. The set of code memories is obtained from an output of the noise encoding module. The noise encoding module is further adapted to sample a latent noisy at an arbitrary time step. The memory-suppress diffusion module 20 comprises three modules, namely a noise encoding module, a normality memorizing module, and a denoise memory-suppress sampling module. The noise encoding module carries out a noise encoding step 24 which follows the basic diffusion process in DDPM to generate a sequence of noise-perturbed images [18]. The noise encoding module thus does not have sufficient structure or material.).
(ii) “normality memorizing module” (Paragraph [0013, 0017, 0018, 0020, 0021 and 0085]-In some embodiments, the memory-suppress diffusion network module further includes a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image, a normality memorizing module adapted to integrate a set of code memories to establish consistent representations of normality, and a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques. In some embodiments, the normality memorizing module is adapted to transform a feature vector associated with one said noise-perturbed image using a corresponding one of the code memories. In some embodiments, during the transforming, the normality memorizing module is further adapted to compute a cosine similarity between the feature vector and the corresponding one of the code memories. In some embodiments, the normality memorizing module is adapted to transform all the feature vectors associated with the plurality of noise-perturbed images to obtain a feature map. In some embodiments, the normality memorizing module is adapted to update a memory query using a feature map. The memory-suppress diffusion module 20 comprises three modules, namely a noise encoding module, a normality memorizing module, and a denoise memory-suppress sampling module. The noise encoding module carries out a noise encoding step 24 which follows the basic diffusion process in DDPM to generate a sequence of noise-perturbed images [18]. Next, a set of code memories obtained from the previous step is integrated to establish consistent representations of normality in a normality memorizing step 26 carried out by the normality memorizing module. The normality memorizing module thus does not have sufficient structure or material.).
(iii) “denoise memory-suppress sampling module” (Paragraph [0013, 0022]- In some embodiments, the memory-suppress diffusion network module further includes a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image, a normality memorizing module adapted to integrate a set of code memories to establish consistent representations of normality, and a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques. In some embodiments, the denoise memory-suppress sampling module is adapted to reconstruct the reconstructed image using knowledge of all previous gradients. The memory-suppress diffusion module 20 comprises three modules, namely a noise encoding module, a normality memorizing module, and a denoise memory-suppress sampling module. The noise encoding module carries out a noise encoding step 24 which follows the basic diffusion process in DDPM to generate a sequence of noise-perturbed images [18]. Next, a set of code memories obtained from the previous step is integrated to establish consistent representations of normality in a normality memorizing step 26 carried out by the normality memorizing module. Lastly, the denoise memory-suppress sampling module carries out a denoise memory-suppress sampling step 28 which continuously reconstructs the normal railway images from the Gaussian noise input using memory-suppression techniques. The denoise memory-suppress sampling module thus does not have sufficient structure or material.).
(iv) “fusion block” (Paragraph [0013, 0022]- In some embodiments, the memory-suppress diffusion network module further includes a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image, a normality memorizing module adapted to integrate a set of code memories to establish consistent representations of normality, and a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques. In some embodiments, the denoise memory-suppress sampling module is adapted to reconstruct the reconstructed image using knowledge of all previous gradients. The memory-suppress diffusion module 20 comprises three modules, namely a noise encoding module, a normality memorizing module, and a denoise memory-suppress sampling module. The noise encoding module carries out a noise encoding step 24 which follows the basic diffusion process in DDPM to generate a sequence of noise-perturbed images [18]. Next, a set of code memories obtained from the previous step is integrated to establish consistent representations of normality in a normality memorizing step 26 carried out by the normality memorizing module. Lastly, the denoise memory-suppress sampling module carries out a denoise memory-suppress sampling step 28 which continuously reconstructs the normal railway images from the Gaussian noise input using memory-suppression techniques. The denoise memory-suppress sampling module thus does not have sufficient structure or material.).
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claim 2, 5-7, and 9-11 along with their dependent claims are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor, or for pre-AIA the applicant regards as the invention.
Claim 2, 5-7, 9-11limitations:
Claim 2; recites the limitation, “a noise encoding module adapted to…..” [Line 3].
Claim 2; recites the limitation, “a normality memorizing module adapted to…...” [Line 5].
Claim 2; recites the limitation, “a denoise memory-suppress sampling module adapted to……,” [Line 8].
Claim 5; recites the limitation, “the noise encoding module is further adapted to…..” [Line 1-2].
Claim 6; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 7; recites the limitation, “the normality memorizing module is further adapted to…...” [Line 1-2].
Claim 9; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 10; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 11; recites the limitation, “the denoise memory-suppress sampling module is adapted to……,” [Line 1-2].
Claim 11; recites the limitation, “a fusion block adapted to……,” [Line 6].
Claims 2, 5-7, 9-11 respectively invokes 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. However, the written description fails to disclose the corresponding structure, material, or acts for performing the entire claimed function and to clearly link the structure, material, or acts to the function. The specification is devoid of adequate structure to perform the claimed functions. The specification does not provide sufficient details such that one of the ordinary skill in the art would understand which structure performed(s) the claimed function.
Therefore, the claim is indefinite and is rejected under 35 U.S.C. 112(b) or pre-AIA 35 U.S.C. 112, second paragraph.
Applicant may:
(a) Amend the claim so that the claim limitation will no longer be interpreted as a limitation under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph;
(b) Amend the written description of the specification such that it expressly recites what structure, material, or acts perform the entire claimed function, without introducing any new matter (35 U.S.C. 132(a)); or
(c) Amend the written description of the specification such that it clearly links the structure, material, or acts disclosed therein to the function recited in the claim, without introducing any new matter (35 U.S.C. 132(a)).
If applicant is of the opinion that the written description of the specification already implicitly or inherently discloses the corresponding structure, material, or acts and clearly links them to the function so that one of ordinary skill in the art would recognize what structure, material, or acts perform the claimed function, applicant should clarify the record by either:
(a) Amending the written description of the specification such that it expressly recites the corresponding structure, material, or acts for performing the claimed function and clearly links or associates the structure, material, or acts to the claimed function, without introducing any new matter (35 U.S.C. 132(a)); or
(b) Stating on the record what the corresponding structure, material, or acts, which are implicitly or inherently set forth in the written description of the specification, perform the claimed function. For more information, see 37 CFR 1.75(d) and MPEP §§ 608.01(o) and 2181.
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 2, 5-7, and 9-11 along with their dependent claims are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for pre-AIA the inventor(s), at the time the application was filed, had possession of the claimed invention. As described above, the disclosure does not provide adequate structure to perform the claimed function in the recited limitation.
Claim 2; recites the limitation, “a noise encoding module adapted to…..” [Line 3].
Claim 2; recites the limitation, “a normality memorizing module adapted to…...” [Line 5].
Claim 2; recites the limitation, “a denoise memory-suppress sampling module adapted to……,” [Line 8].
Claim 5; recites the limitation, “the noise encoding module is further adapted to…..” [Line 1-2].
Claim 6; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 7; recites the limitation, “the normality memorizing module is further adapted to…...” [Line 1-2].
Claim 9; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 10; recites the limitation, “the normality memorizing module is adapted to…...” [Line 1-2].
Claim 11; recites the limitation, “the denoise memory-suppress sampling module is adapted to……,” [Line 1-2].
Claim 11; recites the limitation, “a fusion block adapted to……,” [Line 6].
The specification does not demonstrate that applicant has made an invention that achieves the claimed function because the invention is not described with sufficient detail such that one of ordinary skill in the art can reasonably conclude that the inventor had possession of the claimed invention.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 15, 17-18, and 22-23 are rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB.
Regarding claim 1, MUJKIC explicitly teaches a computer-implemented system (Fig. 1. Paragraph [0026]-MUJKIC discloses the invention provides computer software comprising computer readable instructions which, when executed by one or more electronic processors, causes performance of a method in accordance with any aspect described herein.) for foreign object detection in a scene (Fig. 3. Paragraph [0042]-MUJKIC discloses identification of an obstacle/anomaly within the working environment.); the system comprising:
a) a memory-suppress diffusion network module (Fig. 3, illustrates the use of the network to encode and reconstruct images. Paragraph [0050]) adapted to reconstruct a reconstructed image from an encoded image (Fig. 3. Paragraph [0050]-MUJKIC discloses the input image data is analyzed utilizing an autoencoder method whereby the image data is first encoded utilizing an encoder network to map the image data to a lower-dimensional feature space before subsequently decoding the encoded data utilizing a decoder network to form a reconstructed image of the working environment (step 106).);
the encoded image based on an input image (Fig. 1. Paragraph [0045]-MUJKIC discloses the processor 4 is operable to encode the image data utilizing an encoder network to map the image data to a lower-dimensional feature space.); and
b) a contrastive dissimilarity network (Fig. 3, illustrates a contrastive dissimilarity network. Paragraph [0050]) adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image (Fig. 3, illustrates combining the input image with the reconstructed image at step 108 and an anomaly map output at step 110. Paragraph [0050]- MUJKIC discloses at step 108, the reconstructed image is compared with the input image, here specifically by determined a measure of a perceptual loss between the images. In the illustrated embodiment, an anomaly map is then generated based on the determined perceptual loss.);
Although MUJKIC teaches the memory-suppress diffusion network module trained using normal, real images, MUJKIC fails to explicitly teach wherein the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images.
However, ALREGIB explicitly teaches wherein the memory-suppress diffusion network module (Fig. 1A, illustrates network 100. Paragraph [0024]-ALREGIB discloses the network 100 has been trained with the digit ‘0’ 112, but not the digit ‘5’. The autoencoder needs larger updates to accurately reconstruct the abnormal image, which in this example is the digit ‘5’ 114, than the normal image, digit ‘0’ 112.) and the contrastive dissimilarity network (Fig. 1A, #116 called the gradients. Paragraph [0024]-ALREGIB discloses the gradients 116 indicate the magnitude of the updates that would be necessary to reconstruct the test image (i.e., the digit ‘5’). Therefore, the gradients 116 can be utilized as representations to characterize abnormality of data. One can detect anomalies by measuring how much model update is required by the input compared to normal data (wherein the gradients are indicated in a separate neural network.). Further see annotated Fig. 1A below.) are trained using only normal, real images (Fig. 2. Paragraph [0026]-ALREGIB discloses the training data and the test data set will consist of image data, which can include such image data as: photographic data, video data, point cloud data, and multidimensional data. Image data sometimes includes distortions.).
PNG
media_image1.png
488
739
media_image1.png
Greyscale
Annotated diagram of ALREGIB’s Fig. 1A illustrating two networks, indicated by the two boxes.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of ALREGIB of wherein the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images.
Wherein having MUJKIC’s anomaly detection system wherein the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and ALREGIB relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while ALREGIB there is a need for an anomaly detection system using gradient-based representations that outperforms existing activation-based representation systems. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and ALREGIB et al. (US 20220327389 A1), Paragraph [0013].
Regarding claim 15, MUJKIC in view of ALREGIB explicitly teach the computer-implemented system of claim 1,
MUJKIC further explicitly teaches wherein the system is adapted to provide a weighted dissimilarity score to express the foreign object detection at image-level (Fig. 7, illustrates an anomaly score expressing foreign object detection at image level. Paragraph [0066]-MUJKIC discloses the total anomaly score for the image is calculated as the percentage of pixels with an anomaly score above the threshold value (wherein the total anomaly score is a weighted dissimilarity score).).
Regarding claim 17, MUJKIC in view of ALREGIB explicitly teach the computer-implemented system of claim 1,
MUJKIC further explicitly teaches wherein the memory-suppress diffusion module and the contrastive dissimilarity network are jointly optimized during training (Fig. 3. Paragraph [0053]-MUJKIC discloses the training of AE and VQ-VAE requires a dataset with images depicting normal operating conditions. Further in paragraph [0054]-MUJKIC discloses training of DAE, in addition to the dataset with normal images, requires an annotated dataset of anomaly objects at the pixel level (wherein jointly optimized during training is jointly training with normal images).).
Regarding claim 18, MUJKIC explicitly teaches computer-implemented method (Fig. 1. Paragraph [0026]-MUJKIC discloses the invention provides computer software comprising computer readable instructions which, when executed by one or more electronic processors, causes performance of a method in accordance with any aspect described herein.) for detecting an foreign object (Fig. 3. Paragraph [0042]-MUJKIC discloses identification of an obstacle/anomaly within the working environment.), comprising the steps of:
a) encoding an input image to obtain an encoded image (Fig. 1. Paragraph [0045]-MUJKIC discloses the processor 4 is operable to encode the image data utilizing an encoder network to map the image data to a lower-dimensional feature space.);
b) reconstructing a reconstructed image from the encoded image (Fig. 3. Paragraph [0050]-MUJKIC discloses the input image data is analyzed utilizing an autoencoder method whereby the image data is first encoded utilizing an encoder network to map the image data to a lower-dimensional feature space before subsequently decoding the encoded data utilizing a decoder network to form a reconstructed image of the working environment (step 106).) using a memory- suppress diffusion network module (Fig. 3, illustrates the use of the network to encode and reconstruct images. Paragraph [0050]); and
c) combing the input image and the reconstructed image to predict an anomaly map for the input image (Fig. 3, illustrates combining the input image with the reconstructed image at step 108 and an anomaly map output at step 110. Paragraph [0050]- MUJKIC discloses at step 108, the reconstructed image is compared with the input image, here specifically by determined a measure of a perceptual loss between the images. In the illustrated embodiment, an anomaly map is then generated based on the determined perceptual loss.);
Although MUJKIC teaches the memory-suppress diffusion network module trained using normal, real images, MUJKIC fails to explicitly teach wherein the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images.
However, ALREGIB explicitly teaches wherein the memory-suppress diffusion network module (Fig. 1A, illustrates network 100. Paragraph [0024]-ALREGIB discloses the network 100 has been trained with the digit ‘0’ 112, but not the digit ‘5’. The autoencoder needs larger updates to accurately reconstruct the abnormal image, which in this example is the digit ‘5’ 114, than the normal image, digit ‘0’ 112.) and the contrastive dissimilarity network (Fig. 1A, #116 called the gradients. Paragraph [0024]-ALREGIB discloses the gradients 116 indicate the magnitude of the updates that would be necessary to reconstruct the test image (i.e., the digit ‘5’). Therefore, the gradients 116 can be utilized as representations to characterize abnormality of data. One can detect anomalies by measuring how much model update is required by the input compared to normal data (wherein the gradients are indicated in a separate neural network.). Further see annotated Fig. 1A below.) are trained using only normal, real images (Fig. 2. Paragraph [0026]-ALREGIB discloses the training data and the test data set will consist of image data, which can include such image data as: photographic data, video data, point cloud data, and multidimensional data. Image data sometimes includes distortions.).
PNG
media_image1.png
488
739
media_image1.png
Greyscale
Annotated diagram of ALREGIB’s Fig. 1A illustrating two networks, indicated by the two boxes.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC of a computer-implemented method for detecting an foreign object, comprising the steps of: a) encoding an input image to obtain an encoded image; b) reconstructing a reconstructed image from the encoded image using a memory- suppress diffusion network module; and c) combing the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of ALREGIB of wherein the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images.
Wherein having MUJKIC’s anomaly detection system wherein the memory-suppress diffusion network module and the contrastive dissimilarity network are trained using only normal, real images.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and ALREGIB relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while ALREGIB there is a need for an anomaly detection system using gradient-based representations that outperforms existing activation-based representation systems. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and ALREGIB et al. (US 20220327389 A1), Paragraph [0013].
Regarding claim 22, MUJKIC in view of ALREGIB explicitly teach the method according to claim 18,
MUJKIC further explicitly teaches a non-transitory computer-readable medium, having stored thereon program instructions that, upon execution by a computing device, cause the computing device to perform (Fig. 1, #12 called memory and #2 called controller. Paragraph [0026-0027]-MUJKIC discloses a further aspect of the invention provides computer software comprising computer readable instructions which, when executed by one or more electronic processors, causes performance of a method in accordance with any aspect described herein. A yet further aspect of the invention provides a computer readable medium having the computer software of the preceding aspect of the invention stored thereon.)
Regarding claim 23, MUJKIC in view of ALREGIB explicitly teach the method according to claim 18,
MUJKIC further explicitly teaches a computing system comprising (Fig. 1. Paragraph [0043]-MUJKIC discloses control system 10 comprises a controller 2 having an electronic processor 4, an electronic input 6 and electronic outputs 8, 10. The processor 4 is operable to access a memory 12 of the controller 2 and execute instructions stored therein to perform the steps and functionality of the present invention discussed herein):
a) one or more processors (Fig. 1, #4 called processor. Paragraph [0043]-MUJKIC discloses control system 10 comprises a controller 2 having an electronic processor 4, an electronic input 6 and electronic outputs 8, 10. The processor 4 is operable to access a memory 12 of the controller 2 and execute instructions stored therein to perform the steps and functionality of the present invention discussed herein.); and
b) memory (Fig. 1, #12 called memory.) containing instructions that, when executed by the one or more processors, cause the computing system to perform (Fig. 1, #12 called memory and #2 called controller. Paragraph [0026-0027]-MUJKIC discloses a further aspect of the invention provides computer software comprising computer readable instructions which, when executed by one or more electronic processors, causes performance of a method in accordance with any aspect described herein. A yet further aspect of the invention provides a computer readable medium having the computer software of the preceding aspect of the invention stored thereon.)
Claims 2, 10-11, and 19-20 are rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of PENG et al. (US 20250272960 A1), hereinafter referenced as PENG.
Regarding claim 2, MUJKIC in view of ALREGIB explicitly teach the computer-implemented system of claim 1,
MUJKIC further explicitly teaches wherein the memory-suppress diffusion network module further comprises (Fig. 3, illustrates the use of the network to encode and reconstruct images. Paragraph [0050]):
d) a normality memorizing module adapted to integrate a set of code memories to establish consistent representations of normality (Fig. 3. Paragraph [0058]-MUJKIC discloses all images from normal training datasets are encoded using the trained VQ-VAE to collect a latent code set, and the probability distribution of this latent code set is estimated using Gated-PixelCNN. At the prediction stage, the model yields two output images. The first image is the reconstructed image decoded from the original latent set. Then, when the latent code of an input image is out of the distribution learned in the second step, Gated-PixelCNN will conduct resampling operations on it (wherein resampling is establishing consistent representations of normality and wherein the latent code is a set of code memories).);
the set of code memories obtained from an output of the noise encoding module (Fig. 3. Paragraph [0058]-MUJKIC discloses all images from normal training datasets are encoded using the trained VQ-VAE to collect a latent code set, and the probability distribution of this latent code set is estimated using Gated-PixelCNN (wherein the latent code set is the set of code memories and the VQ-VAE outputs the code set).); and
MUJKIC in view of ALREGIB fail to explicitly teach c) a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image; e) a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques.
However, PENG explicitly teaches c) a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image (Fig. 5. Paragraph [0121]-PENG discloses the processor 206 is configured to introduce the random noise 506 to at least some of the partitioned features or the normal patch features 504 of the plurality of normal images to generate abnormal features (referred to as pseudo-anomaly patch features 508) (wherein the abnormal features are noise-perturbed images).);
e) a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques (Fig. 5. Paragraph [0116]-PENG discloses the RM 510 is a type of anomaly detection model that operates on a principle of reconstructing input data, such as image features. During the training, the RM 510 is trained to learn a compressed representation (encoding) of normal image features 404 and then use it to reconstruct input data (wherein the normal image features are consistent representations of normality and wherein using the compressed representation is the memory-suppression technique).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of c) a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image; e) a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques.
Wherein having MUJKIC’s anomaly detection system c) a noise encoding module adapted to generate a plurality of noise-perturbed images from the input image; e) a denoise memory-suppress sampling module adapted to reconstruct the reconstructed image from the consistent representations of normality using memory-suppression techniques.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
Regarding claim 10, MUJKIC in view of ALREGIB and further in view of PENG explicitly teach the computer-implemented system of claim 2,
MUJKIC in view of ALREGIB fail to explicitly teach wherein the normality memorizing module is adapted to update a memory query using a feature map.
However, PENG explicitly teaches wherein the normality memorizing module is adapted to update a memory query using a feature map (Fig. 1. Paragraph [0072]-PENG discloses the image encoder takes image data of the image 102 as input. This could be in the form of raw pixel values or feature maps generated by a pre-trained convolutional neural network (CNN) (wherein inputting an image is updating a memory query).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of wherein the normality memorizing module is adapted to update a memory query using a feature map.
Wherein having MUJKIC’s anomaly detection system wherein the normality memorizing module is adapted to update a memory query using a feature map.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
Regarding claim 11, MUJKIC in view of ALREGIB and further in view of PENG explicitly teach the computer-implemented system of claim 2,
MUJKIC in view of ALREGIB fail to explicitly teach wherein the denoise memory-suppress sampling module is adapted to reconstruct the reconstructed image using knowledge of all previous gradients.
However, PENG explicitly teaches wherein the denoise memory-suppress sampling module is adapted to reconstruct the reconstructed image using knowledge of all previous gradients (Fig. 5. Paragraph [0116]-PENG discloses the RM 510 is trained to learn a compressed representation (encoding) of normal image features 404 and then use it to reconstruct input data. Further in paragraph [0142]-PENG discloses the features 806 of the image 102 is provided to the RM 510 to perform reconstruction-based AD. The RM 510 Π(.) is a transformer trained to reconstruct the features 806 extracted from the image 102, I, to generate reconstructed image patches 808. The reconstructed image patches 808 are generated based on the pre-trained image encoder 212 of L layers.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of wherein the denoise memory-suppress sampling module is adapted to reconstruct the reconstructed image using knowledge of all previous gradients.
Wherein having MUJKIC’s anomaly detection system wherein the denoise memory-suppress sampling module is adapted to reconstruct the reconstructed image using knowledge of all previous gradients.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
Regarding claim 19, MUJKIC in view of ALREGIB explicitly teach the computer-implemented method of claim 18,
MUJKIC in view of ALREGIB fail to explicitly teach wherein Step a) further comprises a step of generating a plurality of noise-perturbed images from the input image.
However, PENG explicitly teaches wherein Step a) further comprises a step of generating a plurality of noise-perturbed images from the input image (Fig. 5. Paragraph [0121]-PENG discloses the processor 206 is configured to introduce the random noise 506 to at least some of the partitioned features or the normal patch features 504 of the plurality of normal images to generate abnormal features (referred to as pseudo-anomaly patch features 508) (wherein the abnormal features are noise-perturbed images).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented method for detecting an foreign object, comprising the steps of: a) encoding an input image to obtain an encoded image; b) reconstructing a reconstructed image from the encoded image using a memory- suppress diffusion network module; and c) combing the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of wherein Step a) further comprises a step of generating a plurality of noise-perturbed images from the input image.
Wherein having MUJKIC’s anomaly detection system wherein Step a) further comprises a step of generating a plurality of noise-perturbed images from the input image.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
Regarding claim 20, MUJKIC in view of ALREGIB explicitly teach the computer-implemented method of claim 18,
MUJKIC further explicitly teaches wherein Step b) further comprises steps of (Fig. 3. Paragraph [0050]-MUJKIC discloses the input image data is analyzed utilizing an autoencoder method whereby the image data is first encoded utilizing an encoder network to map the image data to a lower-dimensional feature space before subsequently decoding the encoded data utilizing a decoder network to form a reconstructed image of the working environment (step 106).):
d) integrating a set of code memories to establish consistent representations of normality (Fig. 3. Paragraph [0058]-MUJKIC discloses all images from normal training datasets are encoded using the trained VQ-VAE to collect a latent code set, and the probability distribution of this latent code set is estimated using Gated-PixelCNN. At the prediction stage, the model yields two output images. The first image is the reconstructed image decoded from the original latent set. Then, when the latent code of an input image is out of the distribution learned in the second step, Gated-PixelCNN will conduct resampling operations on it (wherein resampling is establishing consistent representations of normality and wherein the latent code is a set of code memories).);
the set of code memories obtained from an output of Step a) (Fig. 3. Paragraph [0058]-MUJKIC discloses all images from normal training datasets are encoded using the trained VQ-VAE to collect a latent code set, and the probability distribution of this latent code set is estimated using Gated-PixelCNN (wherein the latent code set is the set of code memories and the VQ-VAE outputs the code set).); and
MUJKIC in view of ALREGIB fail to explicitly teach e) reconstructing the reconstructed image from the consistent representations of normality using memory-suppression techniques
However, PENG explicitly teaches e) reconstructing the reconstructed image from the consistent representations of normality using memory-suppression techniques (Fig. 5. Paragraph [0116]-PENG discloses the RM 510 is a type of anomaly detection model that operates on a principle of reconstructing input data, such as image features. During the training, the RM 510 is trained to learn a compressed representation (encoding) of normal image features 404 and then use it to reconstruct input data (wherein the normal image features are consistent representations of normality and wherein using the compressed representation is the memory-suppression technique).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented method for detecting an foreign object, comprising the steps of: a) encoding an input image to obtain an encoded image; b) reconstructing a reconstructed image from the encoded image using a memory- suppress diffusion network module; and c) combing the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of e) reconstructing the reconstructed image from the consistent representations of normality using memory-suppression techniques.
Wherein having MUJKIC’s anomaly detection system e) reconstructing the reconstructed image from the consistent representations of normality using memory-suppression techniques.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
Claims 3-5 are rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of PENG et al. (US 20250272960 A1), hereinafter referenced as PENG, and further in view of RHODES et al. (US 20240144447 A1), hereinafter referenced as RHODES.
Regarding claim 3, MUJKIC in view of ALREGIB and further in view of PENG explicitly teach the computer-implemented system of claim 2,
MUJKIC in view of ALREGIB and further in view of PENG fail to explicitly teach wherein the plurality of noise-perturbed images is generated with a steadily increasing noise level.
However, RHODES explicitly teaches wherein the plurality of noise-perturbed images is generated with a steadily increasing noise level (Fig. 1. Paragraph [0021]-RHODES discloses a corrupting operation at index t takes the output of the corrupting operation at index t−1 as input and adds, e.g., Gaussian noise, to the input. x.sub.0˜q(x.sub.0) may represent an initial (uncorrupted) image, and x.sub.t may represent a corrupted image following t corrupting operations of the forward diffusion process 102. x.sub.T may represent a corrupted image following T corrupting operations. The forward diffusion process 102 can form a Markov Process that may gradually add Gaussian noise according to a variance schedule corresponding to the different indices. The corrupting operations of the forward diffusion process 102 may receive an uncorrupted input and produce progressively more noisy images at each index.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of RHODES of wherein the plurality of noise-perturbed images is generated with a steadily increasing noise level.
Wherein having MUJKIC’s anomaly detection system wherein the plurality of noise-perturbed images is generated with a steadily increasing noise level.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and RHODES relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while RHODES information can be used to provide useful insights for improving methods aimed at detecting and protecting against the malicious use of synthetic data (e.g., deep fakes), adversarial attacks, building trust in deep learning models, etc. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and RHODES et al. (US 20240144447 A1), Paragraph [0018].
Regarding claim 4, MUJKIC in view of ALREGIB and further in view of PENG explicitly teach the computer-implemented system of claim 2,
MUJKIC in view of ALREGIB and further in view of PENG fail to explicitly teach wherein the noise levels of the plurality of noise-perturbed images follow a Markovian process, and sizes of steps of the noise levels are dominated by a variance scheduler.
However, RHODES explicitly teaches wherein the noise levels of the plurality of noise-perturbed images follow a Markovian process, and sizes of steps of the noise levels are dominated by a variance scheduler (Fig. 1. Paragraph [0021]-RHODES discloses a corrupting operation at index t takes the output of the corrupting operation at index t−1 as input and adds, e.g., Gaussian noise, to the input. x.sub.0˜q(x.sub.0) may represent an initial (uncorrupted) image, and x.sub.t may represent a corrupted image following t corrupting operations of the forward diffusion process 102. x.sub.T may represent a corrupted image following T corrupting operations. The forward diffusion process 102 can form a Markov Process that may gradually add Gaussian noise according to a variance schedule corresponding to the different indices. The corrupting operations of the forward diffusion process 102 may receive an uncorrupted input and produce progressively more noisy images at each index.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of RHODES of wherein the noise levels of the plurality of noise-perturbed images follow a Markovian process, and sizes of steps of the noise levels are dominated by a variance scheduler.
Wherein having MUJKIC’s anomaly detection system wherein the noise levels of the plurality of noise-perturbed images follow a Markovian process, and sizes of steps of the noise levels are dominated by a variance scheduler.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and RHODES relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while RHODES information can be used to provide useful insights for improving methods aimed at detecting and protecting against the malicious use of synthetic data (e.g., deep fakes), adversarial attacks, building trust in deep learning models, etc. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and RHODES et al. (US 20240144447 A1), Paragraph [0018].
Regarding claim 5, MUJKIC in view of ALREGIB and further in view of PENG explicitly teach the computer-implemented system of claim 2,
MUJKIC in view of ALREGIB and further in view of PENG fail to explicitly teach wherein the noise encoding module is further adapted to sample a latent noisy at an arbitrary time step.
However, RHODES explicitly teaches wherein the noise encoding module is further adapted to sample a latent noisy at an arbitrary time step (Fig. 2. Paragraph [0046]-RHODES discloses visualizer 204 may extract information about the denoising process of sampling network 106, e.g., from the noise values predicted by the denoising operations 110 at different indices.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of RHODES of wherein the noise encoding module is further adapted to sample a latent noisy at an arbitrary time step.
Wherein having MUJKIC’s anomaly detection system wherein the noise encoding module is further adapted to sample a latent noisy at an arbitrary time step.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and RHODES relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while RHODES information can be used to provide useful insights for improving methods aimed at detecting and protecting against the malicious use of synthetic data (e.g., deep fakes), adversarial attacks, building trust in deep learning models, etc. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and RHODES et al. (US 20240144447 A1), Paragraph [0018].
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of PENG et al. (US 20250272960 A1), hereinafter referenced as PENG, and further in view of NOH et al. (US 20230410249 A1), hereinafter referenced as NOH.
Regarding claim 6, MUJKIC in view of ALREGIB and further in view of PENG explicitly teach the computer-implemented system of claim 2,
MUJKIC in view of ALREGIB and further in view of PENG fail to explicitly teach wherein the normality memorizing module is adapted to transform a feature vector associated with one said noise-perturbed image using a corresponding one of the code memories.
However, NOH explicitly teaches wherein the normality memorizing module is adapted to transform a feature vector associated with one said noise-perturbed image using a corresponding one of the code memories (Fig. 4. Paragraph [0082]-NOH discloses an image processing method performed by the morphing generator 410 may include extracting content latent codes c.sub.A and c.sub.B and style latent codes s.sub.A and s.sub.B from each of a plurality of input images x.sub.AA and x.sub.BB, obtaining a content feature vector c.sub.α by calculating a weighted sum of the content latent codes c.sub.A and c.sub.B extracted from the input images based on a morphing control parameter a, obtaining a style feature vector s.sub.α by calculating a weighted sum of the style latent codes S.sub.A and s.sub.B extracted from the input images based on the morphing control parameter α, and generating a morphing image y.sub.αα based on the content feature vector c.sub.α and the style feature vector s.sub.α. (wherein an input image is input into an autoencoder with causes the image to be noise-perturbed and wherein transforming is generating a morphing image).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of NOH of wherein the normality memorizing module is adapted to transform a feature vector associated with one said noise-perturbed image using a corresponding one of the code memories.
Wherein having MUJKIC’s anomaly detection system wherein the normality memorizing module is adapted to transform a feature vector associated with one said noise-perturbed image using a corresponding one of the code memories.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and NOH relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while NOH to allow a user to separately control a degree of a content transition related to a composition of an image and a degree of a style transition related to an appearance of an image to obtain a morphing image between images, and thus provide rich usability. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and NOH et al. (US 20230410249 A1), Paragraph [0130].
Claims 7-9 are rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of PENG et al. (US 20250272960 A1), hereinafter referenced as PENG, and further in view of NOH et al. (US 20230410249 A1), hereinafter referenced as NOH, and further in view of ZHANG et al. (US 20250209783 A1), hereinafter referenced as ZHANG.
Regarding claim 7, MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH explicitly teach the computer-implemented system of claim 6,
MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH fail to explicitly teach wherein during the transforming, the normality memorizing module is further adapted to compute a cosine similarity between the feature vector and the corresponding one of the code memories.
However, ZHANG explicitly teaches wherein during the transforming, the normality memorizing module is further adapted to compute a cosine similarity between the feature vector and the corresponding one of the code memories (Fig. 1. Paragraph [0055]-ZHANG discloses the server may first determine a feature similarity between the first image sample features corresponding to the remaining M−1 first image samples in the M first image samples and the first image sample corresponding to the j.sup.th first image sample (wherein the first image). Further in paragraph [0055]-ZHANG discloses the feature similarity may be a cosine similarity. For example, first image sample features of the same batch of image samples are first pooled into a sample feature vector v. Then, the cosine similarity between image samples in the same batch may be determined by using the following formula:
PNG
media_image2.png
90
269
media_image2.png
Greyscale
where S represents the cosine similarity between image samples in the same batch, Softmax represents a normalization function, v represents the sample feature vector v determined by pooling the first image sample features of the same batch of image samples, v.sup.T represents a transposed matrix of v, and ∥v∥.sub.2 represents a two-norm of v (wherein the first image sample features corresponding to the remaining M−1 first image samples in the M first image samples is the feature vector and wherein first image sample corresponding to the j.sup.th first image sample is the corresponding on of the code memories).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of ZHANG of wherein during the transforming, the normality memorizing module is further adapted to compute a cosine similarity between the feature vector and the corresponding one of the code memories.
Wherein having MUJKIC’s anomaly detection system wherein during the transforming, the normality memorizing module is further adapted to compute a cosine similarity between the feature vector and the corresponding one of the code memories.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and ZHANG relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while ZHANG the identification model has a good generalization capability for images having different scenes and images generated in different modes, thereby effectively improving model identification accuracy. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and ZHANG et al. (US 20250209783 A1), Paragraph [0022].
Regarding claim 8, MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH and further in view of ZHANG explicitly teach the computer-implemented system of claim 7,
MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH fail to explicitly teach wherein a Softmax function is used to obtains weights in computation of the cosine similarity.
However, ZHANG explicitly teaches wherein a Softmax function is used to obtains weights in computation of the cosine similarity (Fig. 1. Paragraph [0055]-ZHANG discloses the feature similarity may be a cosine similarity. For example, first image sample features of the same batch of image samples are first pooled into a sample feature vector v. Then, the cosine similarity between image samples in the same batch may be determined by using the following formula:
PNG
media_image2.png
90
269
media_image2.png
Greyscale
where S represents the cosine similarity between image samples in the same batch, Softmax represents a normalization function, v represents the sample feature vector v determined by pooling the first image sample features of the same batch of image samples, v.sup.T represents a transposed matrix of v, and ∥v∥.sub.2 represents a two-norm of v.).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of ZHANG of wherein a Softmax function is used to obtains weights in computation of the cosine similarity.
Wherein having MUJKIC’s anomaly detection system wherein a Softmax function is used to obtains weights in computation of the cosine similarity.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and ZHANG relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while ZHANG the identification model has a good generalization capability for images having different scenes and images generated in different modes, thereby effectively improving model identification accuracy. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and ZHANG et al. (US 20250209783 A1), Paragraph [0022].
Regarding claim 9, MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH explicitly teach the computer-implemented system of claim 6,
MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH fail to explicitly teach wherein the normality memorizing module is adapted to transform all the feature vectors associated with the plurality of noise-perturbed images to obtain a feature map.
However, ZHANG explicitly teaches wherein the normality memorizing module is adapted to transform all the feature vectors associated with the plurality of noise-perturbed images to obtain a feature map (Fig. 1. Paragraph [0100]-ZHANG discloses S31: The server maps the first image sample feature to a feature space of the first texture image, to obtain a mapped sample feature (wherein a mapped sample feature is a feature map and wherein the first image sample feature is the feature vector).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of PENG and further in view of NOH of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of ZHANG of wherein the normality memorizing module is adapted to transform all the feature vectors associated with the plurality of noise-perturbed images to obtain a feature map.
Wherein having MUJKIC’s anomaly detection system wherein the normality memorizing module is adapted to transform all the feature vectors associated with the plurality of noise-perturbed images to obtain a feature map.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and ZHANG relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while ZHANG the identification model has a good generalization capability for images having different scenes and images generated in different modes, thereby effectively improving model identification accuracy. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and ZHANG et al. (US 20250209783 A1), Paragraph [0022].
Claims 12 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of LIU et al. (US 20240185388 A1), hereinafter referenced as LIU, and further in view of PENG et al. (US 20250272960 A1), hereinafter referenced as PENG, and further in view of YU et al. (US 20210319420 A1), hereinafter referenced as YU.
Regarding claim 12, MUJKIC in view of ALREGIB explicitly teach the computer-implemented system of claim 1,
MUJKIC further explicitly teaches wherein the contrastive dissimilarity network comprises (Fig. 3, illustrates a contrastive dissimilarity network. Paragraph [0050]):
Although MUJKIC explicitly teaches an encoder adapted to encode the input image and the reconstructed image, MUJKIC in view of ALREGIB fail to explicitly teach f) an encoder adapted to encode the input image and the reconstructed image to obtain two embedding vectors.
However, LIU explicitly teaches f) an encoder adapted to encode the input image and the reconstructed image to obtain two embedding vectors (Fig. 5. Paragraph [0063]-LIU discloses during the training process, for the given hidden vector, preferred reconstructed sample data is generated by optimizing the decoder (wherein the hidden vectors are embedding vectors). Further in paragraph [0065]-LIU discloses input data 501 containing original image 202, low-resolution image 203, and reference image 201 may be first input to image encoder 502-1 for encoding, and the encoded data may then be upsampled via sampler 503 to generate sample 504. Sample 504 may be input to image decoder 505 for generating a reconstructed image. The reconstructed image is then input to image encoder 502-2, and image encoder 502-2 compares the received reconstructed image with the original image to determine whether the reconstructed image meets a predetermined condition.);
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of LIU of f) an encoder adapted to encode the input image and the reconstructed image to obtain two embedding vectors.
Wherein having MUJKIC’s anomaly detection system having f) an encoder adapted to encode the input image and the reconstructed image to obtain two embedding vectors.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and LIU relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while LIU two training processes are performed iteratively. This ultimately improves the ability of image encoder 502-1 or 502-2 to discriminate the reconstructed image and the ability of image decoder 505 to generate a reconstructed image that is as similar as possible to the original image. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and LIU et al. (US 20240185388 A1), Paragraph [0066].
MUJKIC in view of ALREGIB and further in view of LIU fail to explicitly teach g) a projector adapted to project the two embedding vectors to a larger space.
However, PENG explicitly teaches g) a projector (Fig. 1, #114 called projector operator. Paragraph [0074]) adapted to project the two embedding vectors to a larger space (Fig. 1. Paragraph [0074]-PENG discloses the system 106 is configured to project each of the feature patches 104 into the latent space 116 using a projector operator 114 (wherein the feature patches are embedding vectors).); and
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of LIU of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of g) a projector adapted to project the two embedding vectors to a larger space.
Wherein having MUJKIC’s anomaly detection system having g) a projector adapted to project the two embedding vectors to a larger space.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG fail to explicitly teach h) a fusion block adapted to compute a correlation map from an output of the projector.
However, YU explicitly teaches h) a fusion block adapted to compute a correlation map from an output of the projector (Fig. 5. Paragraph [0077]-YU discloses then the generated correlation feature maps are fed into fusion block 572 and fusion block 582 respectively, where the feature maps with different sizes are aligned in both spatial and channel domains. For example, the features with low-resolution are up-sampled, while the features with high-resolution are down-sampled to the same scale. Then a 1×1 convolution is applied to the spatially aligned features to further align them in channel dimension. Finally, all aligned features are subsequently fused by an element-wise sum operation (wherein the generated feature maps are an output of the projector).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of YU of h) a fusion block adapted to compute a correlation map from an output of the projector.
Wherein having MUJKIC’s anomaly detection system having h) a fusion block adapted to compute a correlation map from an output of the projector.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and YU relate to detecting data in images and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while YU is to improve a computing system's precision and robustness for tracking an object with an arbitrary shape, e.g., with a bounding box or an accurate mask of the object. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and YU et al. (US 20210319420 A1), Paragraph [0008].
Regarding claim 21, MUJKIC in view of ALREGIB explicitly teach the computer-implemented system of claim 18,
MUJKIC further explicitly teaches wherein Step c) further comprises steps of (Fig. 3, illustrates combining the input image with the reconstructed image at step 108 and an anomaly map output at step 110. Paragraph [0050]- MUJKIC discloses at step 108, the reconstructed image is compared with the input image, here specifically by determined a measure of a perceptual loss between the images. In the illustrated embodiment, an anomaly map is then generated based on the determined perceptual loss.):
MUJKIC in view of ALREGIB fail to explicitly teach f) encoding the input image and the reconstructed image to obtain two embedding vectors.
However, LIU explicitly teaches f) encoding the input image and the reconstructed image to obtain two embedding vectors (Fig. 5. Paragraph [0063]-LIU discloses during the training process, for the given hidden vector, preferred reconstructed sample data is generated by optimizing the decoder (wherein the hidden vectors are embedding vectors). Further in paragraph [0065]-LIU discloses input data 501 containing original image 202, low-resolution image 203, and reference image 201 may be first input to image encoder 502-1 for encoding, and the encoded data may then be upsampled via sampler 503 to generate sample 504. Sample 504 may be input to image decoder 505 for generating a reconstructed image. The reconstructed image is then input to image encoder 502-2, and image encoder 502-2 compares the received reconstructed image with the original image to determine whether the reconstructed image meets a predetermined condition.);
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented method for detecting an foreign object, comprising the steps of: a) encoding an input image to obtain an encoded image; b) reconstructing a reconstructed image from the encoded image using a memory- suppress diffusion network module; and c) combing the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of LIU of f) encoding the input image and the reconstructed image to obtain two embedding vectors.
Wherein having MUJKIC’s anomaly detection system having f) encoding the input image and the reconstructed image to obtain two embedding vectors.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and LIU relate to encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while LIU two training processes are performed iteratively. This ultimately improves the ability of image encoder 502-1 or 502-2 to discriminate the reconstructed image and the ability of image decoder 505 to generate a reconstructed image that is as similar as possible to the original image. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and LIU et al. (US 20240185388 A1), Paragraph [0066].
MUJKIC in view of ALREGIB and further in view of LIU fail to explicitly teach g) projecting the two embedding vectors to a larger space; and
However, PENG explicitly teaches g) projecting the two embedding vectors to a larger space (Fig. 1. Paragraph [0074]-PENG discloses the system 106 is configured to project each of the feature patches 104 into the latent space 116 using a projector operator 114 (wherein the feature patches are embedding vectors).); and.
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of LIU of a computer-implemented method for detecting an foreign object, comprising the steps of: a) encoding an input image to obtain an encoded image; b) reconstructing a reconstructed image from the encoded image using a memory- suppress diffusion network module; and c) combing the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of PENG of g) projecting the two embedding vectors to a larger space; and.
Wherein having MUJKIC’s anomaly detection system having g) projecting the two embedding vectors to a larger space; and.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and PENG relate to anomaly detection and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while PENG to improve an identification capability of the model for objects in different scene environments, in this application, a second image sample feature is generated based on the first image sample feature. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and PENG et al. (US 20250272960 A1), Paragraph [0022].
MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG fail to explicitly teach h) computing a correlation map from an output of Step g).
However, YU explicitly teaches h) computing a correlation map from an output of Step g) (Fig. 5. Paragraph [0077]-YU discloses then the generated correlation feature maps are fed into fusion block 572 and fusion block 582 respectively, where the feature maps with different sizes are aligned in both spatial and channel domains. For example, the features with low-resolution are up-sampled, while the features with high-resolution are down-sampled to the same scale. Then a 1×1 convolution is applied to the spatially aligned features to further align them in channel dimension. Finally, all aligned features are subsequently fused by an element-wise sum operation (wherein the generated feature maps are an output of step g).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of YU of h) computing a correlation map from an output of Step g).
Wherein having MUJKIC’s anomaly detection system having h) computing a correlation map from an output of Step g).
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and YU relate to detecting data in images and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while YU is to improve a computing system's precision and robustness for tracking an object with an arbitrary shape, e.g., with a bounding box or an accurate mask of the object. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and YU et al. (US 20210319420 A1), Paragraph [0008].
Claims 13-14 are rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of LIU et al. (US 20240185388 A1), hereinafter referenced as LIU, and further in view of PENG et al. (US 20250272960 A1), hereinafter referenced as PENG, and further in view of YU et al. (US 20210319420 A1), hereinafter referenced as YU, and further in view of AHMADI et al. (US 20230118009 A1), hereinafter referenced as AHMADI.
Regarding claim 13, MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG and further in view of YU explicitly teach the computer-implemented system of claim 12,
Although MUJKIC explicitly teaches a pre-trained VGG (Visual Geometry Group) model, MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG and further in view of YU fail to explicitly teach wherein the encoder is a pre-trained VGG (Visual Geometry Group) model.
However, AHMADI discloses wherein the encoder is a pre-trained VGG (Visual Geometry Group) model (Fig. 4. Paragraph [0086]-AHMADI discloses an image (e.g., an aerial image) 410 including multiple properties on a piece of land may be received as input to an image segmentation model or module 420 (e.g., a U-Net model with VGG-16 encoder also known as “UNET-VGG”).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG and further in view of YU of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of AHMADI of wherein the encoder is a pre-trained VGG (Visual Geometry Group) model.
Wherein having MUJKIC’s anomaly detection system wherein the encoder is a pre-trained VGG (Visual Geometry Group) model.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and AHMADI relate to detecting anomalies and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while AHMADI the main bottleneck of existing state-of-the-art models is related to separation and split of close-by properties, therefore a correction model, which can be a trained neural network model, is implemented in the embodiments described herein to split the segmented images properly.. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and AHMADI et al. (US 20230118009 A1), Paragraph [0089].
Regarding claim 14, MUJKIC in view of ALREGIB and further in view of LIU and further in view of PENG and further in view of YU and further in view of AHMADI explicitly teach the computer-implemented system of claim 13,
MUJKIC further explicitly teaches wherein the projector is a three-layer perceptron with batch normalization and ReLU activation (Fig. 4, illustrates a multilayer percepton with batch normalization and ReLU activation. Paragraph [0057]-MUJKIC discloses each convolutional layer, with the exception of the final layer, is followed by batch normalization and LeakyReLu as activation function.).
Claim 16 is rejected under 35 U.S.C. 103 as being unpatentable over MUJKIC et al. (US 20230278550 A1), hereinafter referenced as MUJKIC, in view of ALREGIB et al. (US 20220327389 A1), hereinafter referenced as ALREGIB, and further in view of CHEN et al. (US 20250348987 A1), hereinafter referenced as CHEN.
Regarding claim 16, MUJKIC in view of ALREGIB explicitly teach the computer-implemented system of claim 15,
MUJKIC in view of ALREGIB fail to explicitly teach wherein the system is adapted to generate a stacked pixel-wise anomaly map by merging a score distance map and a feature distance map along a depth dimension.
However, CHEN explicitly teaches wherein the system is adapted to generate a stacked pixel-wise anomaly map by merging a score distance map and a feature distance map along a depth dimension (Fig. 6, illustrates merging maps along a depth dimension. Paragraph [0129]-CHEN discloses the RGB feature map and the depth feature map are subjected to feature-splicing, upsampling, and channel merging layer by layer through the feature fusion layer in the feature extraction network 610, so as to obtain a fused feature map; the fused feature map is input into the feature classification network 620 and the feature reconstruction network 630; by performing feature classification pixel by pixel in the feature classification network 620, a classified score map indicating the probability that each pixel point is defective can be obtained (wherein the RGB feature map is a score distance map and the depth feature map is a feature distance map).).
Therefore, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention was made to combine the teachings of MUJKIC in view of ALREGIB of a computer-implemented system for foreign object detection in a scene; the system comprising: a) a memory-suppress diffusion network module adapted to reconstruct a reconstructed image from an encoded image; the encoded image based on an input image; and b) a contrastive dissimilarity network adapted to combine the input image and the reconstructed image to predict an anomaly map for the input image with the teachings of CHEN of wherein the system is adapted to generate a stacked pixel-wise anomaly map by merging a score distance map and a feature distance map along a depth dimension.
Wherein having MUJKIC’s anomaly detection system wherein the system is adapted to generate a stacked pixel-wise anomaly map by merging a score distance map and a feature distance map along a depth dimension.
The motivation behind the modification would have been to obtain an anomaly detection system that enhances the ability of the system to efficiently and correctly identify anomalies. Since both MUJKIC and CHEN relate to detecting anomalies and encoding and reconstructing images, wherein MUJKIC since anomalies deviate from normal data instances, they are more difficult to be reconstructed from the same low-dimensional feature space and have higher reconstruction error. Therefore, the present invention utilizes this reconstruction error to identify anomalies within the environment and optionally generate anomaly maps therefrom, e.g. for assisting an operator of the associated machine, while CHEN provide a defect detection method, which can combine global anomaly detection and local anomaly detection to train a defect detection model, thereby improving defect detection capability of the defect detection model, such that when the defect detection model is used to perform defect detection, large defects as well as small defects in a to-be-detected object can be detected, thereby improving the comprehensiveness and accuracy of defect detection. Please see MUJKIC et al. (US 20230278550 A1), Paragraph [0041], and CHEN et al. (US 20250348987 A1), Paragraph [0036].
Conclusion
Listed below are the prior arts made of record and not relied upon but are considered pertinent to applicant’s disclosure.
BERGMANN et al. (US 20230073223 A1) - This invention relates generally to machine vision systems, and more particularly, to the detection of anomalies in scenes observed by imaging sensors…Abstract, Fig. 3.
TAN et al. (US 20220156910 A1) - A method for generating reconstruction a reconstructed image is adapted to an input image having a target object. The method comprises converting the input image into a feature map with vectors by an encoder; performing a training procedure according to training images of reference objects to generate feature prototypes associated with the training images and store the feature prototypes to a memory; selecting a part of feature prototypes from the feature prototypes stored in the memory according to similarities between the feature prototypes and the feature vectors; generating a similar feature map according the part of feature prototypes and weights, wherein the weights represents similarities between the part of feature prototypes and the feature vectors; and converting the similar feature map into the reconstructed image by a decoder; wherein the encoder, the decoder and the memory form an auto-encoder…Abstract, Fig. 1.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ETHAN N WOLFSON whose telephone number is (571)272-1898. The examiner can normally be reached Monday - Friday 8:00 am - 5:00 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Chineyere Wills-Burns can be reached at (571) 272-9752. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/ETHAN N WOLFSON/Examiner, Art Unit 2673
/CHINEYERE WILLS-BURNS/Supervisory Patent Examiner, Art Unit 2673
Prosecution Insights