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 .
Status of Claims
This Office Action is in response to the application filed on 10/21/2024. Claims 1-20 have been examined.
Priority
Receipt is acknowledged of certified copies of papers submitted under 35 U.S.C. 119(a)- (d), which have been placed of record in the file.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 10/21/2024 filed in accordance with the provisions of 37 CFR 1.97. Accordingly, it is being considered by the examiner.
Specification
The specification has not been reviewed in detail to determine the presence of all possible minor errors. Applicant's attention is directed to the specification, and applicant is requested to make any necessary corrections of which applicant is aware.
Allowable Subject Matter
Claims 5 and corresponding claims 6-11, as well as claim 17, are objected to as being depend upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
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.
The factual inquiries for establishing a background for determining obviousness under 35
U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness
Claims 1-2, 12-14, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over by Zhihao Gu, Yang Chen, Taiping Yao, Shouhong Ding, Jilin Li, Lizhuang Ma (“Delving into the local: Dynamic Inconsistency Learning for DeepFake Video Detection” hereinafter “GU” (A copy of the non-patent literature reference by GU et al (“Gu”) is attached to this office action) in view Hua et al (US 20230222842 A1) hereinafter “Hua”.
Regarding Claim 1 Gu-Hua
Gu discloses 1. A video detection method, performed by an electronic device, (Gu, page 744 and 746 discloses a deepFake video detection method using a neural network architecture including convolution, Intra-Sim, Inter-Sim, and Fake/Real classifier.) comprising:
extracting a plurality of video snippets of a target video; (Gu, page 746 “To this end, we sample the video sequence uniformly into U snip-pets, each of which contains T successive frames rather than a single frame like pervious works do.” See also Figure 2 with 3 snippets)
. . ., the local features being configured for representing a time sequence inconsistency of the video snippets; (Gu, page 746 “Intra-Snippet Inconsistency Module(Intra-SIM) then takes frames with in each snippet to model the local inconsistency encoded in subtle motions….The long branch consists of an Intra Snippet Motion Attention(Intra-SMA) module and a path way with learnable convolutional kernels to adaptatively aggregate intra-snippet inconsistency information….To model temporal relation, Intra-SMA applies bi-directional temporal difference guided coordinate attention to make the network attend to local network.”)
performing fusion operation on the local features to obtain a global feature of the target video; (Gu, page 745 “All snippets span uniformly over the entire video to form a local-to-global view. To deal with these snippets, an Intra-Snippet Inconsistency Module (Intra-SIM) and an Inter-Snippet Interaction Module (Inter-SIM) are devised and they are performed in an alternate manner. Specifically, the Intra-SIM first adopts bi-directional temporal difference operation to model the intra-snippet inconsistency…The Inter-SIM employs a new two-branch structure to establish a cross-snippet view for interaction promotion.”) and
determining an authenticity probability of the target object in the target video based on the global feature, and obtaining a detection result of the target video based on the authenticity probability. (Gu, Fig, 2, page 746 teaches determining whether the target video is Fake/Real based on the global representation generated by the network, Gu, page 750 further teaches successfully localizing fake face and spotting fake frames, thereby obtaining a detection result of the target video.)
PNG
media_image1.png
102
618
media_image1.png
Greyscale
Figure 2: The overall architecture of the proposed method.
Gu does not expressly disclose
extracting local features corresponding to the video snippets, respectively, based on motion information of a target object in the video snippets,..
However, in the same field of endeavor Hua discloses more explicitly the following:
extracting local features corresponding to the video snippets, respectively, based on motion information of a target object in the video snippets, (Hua , [0071] “One feature of the motion feature extraction module 412 is to extract motion features for each pair of time-constrained sequential frames from the time-stamped frame sequence from the video stream, and then use the frames to help characterize and quantify the motion of objects. ”[0084] “The motion features determined in step 432 may be determined, for example, by concatenating all V for each pixel in the applicable ROI, although other mathematical methods for determining an overall vector from a collection of individual vectors may also be used. So the motion feature M.sub.f(i) is the vector of velocity of all pixels in the face region A; the motion feature M.sub.o(i) is the vector of velocity of all pixels in the face surrounding region B; the motion feature M.sub.b(i) is the vector of velocity of all pixels in the background region C. Motion features extracted in the background region are used as reference knowledge for comparing motion features from other two regions.”)
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to modify the teachings of Gu with Kim to configure the system of Gu to perform extracting local features corresponding to the video snippets, respectively, based on motion information of a target object in the video snippets as taught by Hua. Hua teaches that motion information extracted from a video stream provides useful information for distinguishing live faces from spoofed faces by analyzing motion features of the target object.
Accordingly, One of ordinary skill in the art would have been motivated to incorporate Hua’s motion-information-based feature extraction into Gu to improve spoof detection and enhance the robustness and reliability of video-based face authentication system. (Hua ¶¶[0030], [0068]-[0072])
Regarding Claim 2 GU-Hua
Gu-Hua discloses 2. The method according to claim 1,
dividing the target object into a plurality of regions based on the motion information; (Hua, [0079] “Returning to the method 400, at step 416, I.sub.i(x, y, t) is determined for each pixel in the i.sup.th image. At step 420, the i.sup.th frame from a time-stamped frame sequence undergoes segmentation into regions of interest (ROIs). Careful segmentation into various regions of interest is an important step for measuring motion interactions between face and background in terms of optical flow. This relationship is important because it characterizes a diverse set of live and spoof scenarios. [0080] “Referring briefly to FIG. 6, three primary regions of a given image are proposed for measurement and comparison:” See that Fig. 6C)
performing feature extraction on the video snippets, and integrating extracted features based on a time dimension, to obtain a plurality of time sequence convolution features; (Gu discloses performing features extraction on the video snippets, and integrating extracted features based on a time dimension, to obtain a plurality of time sequence convolution feature based, because Gu extracts inconsistency information from sampled video snippets, each comprising successive video frames, using an Intra-Snippet inconsistency information from sampled video snippets, each comprising successive video frames, using an Intra-Snippet Inconsistency Module (Intra-Sim) w. “To deal with these snip pets, an Intra-Snippet Inconsistency Module (Intra-SIM)..” “The intra-Sim performs bi-directional temporal difference operations to model intra-snippet inconsistency and aggregates the extracted temporal information. )”…generates dynamic convolutional kernels to adaptively aggregate long-range temporal information from adjacent snippets ”. Gu further generates a temporal convolution representation (Conv_U) from the temporal features (Gu, pp. 745-748, Intra-SIM discussion and Eq.(11)) and
obtaining the local features of the video snippets based on the plurality of regions and the plurality of time sequence convolution features. (Gu discloses obtaining local snippet representations (i.e., local feature) by modeling local inconsistency within each video snippet and adaptively aggregating the intra-snippet information. Gu further generates a temporal convolution (ConvU) from temporal features across adjacent video snippets. (Gu, pp. 745-746, Fig. 2 Eqs. (10) (11))
Note: The motivation set forth in the rejection of claim 1 applies equally as well to claim 2.
Regarding Claim 12 Gu-Hua
Gu-Hua discloses 12. A non-transitory computer-readable storage medium, storing one or more computer programs that, when executed by one or more processors of an electronic device, cause the electronic device (Hua [0172] “A computer readable storage medium may be any non-transitory, tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, device, computer, computing system, computer system, or any programmable machine or device.”) to perform the method according to claim 1.
Note: The motivation set forth in the rejection of claim 1 applies equally as well to claim 12.
Regarding Claim 13 Gu-Hua
Gu-Hua discloses 13. An electronic device, comprising:
one or more processors; (Hua, Fig. 1, 104 “processor” [0115] “one or more processors (such as the processor 104) a processor,…”) and
one or more storage apparatuses storing one or more programs that, when executed by the one or more processors, cause the electronic device to: (Hua, Fig. 1B, 166 “Face Enrollment Model” [0052] “The face enrollment module 166 may comprise a processor and one or more memory blocks containing, for example, instructions for execution by the processor as well as data received from another module…” [0172] “A computer readable storage medium…can contain or store a program for use by or in connection with an instruction execution system, apparatus, device, computer,…or any programmable machine or device.”)
The remaining limitations of independent claim 13 recite features that are substantially similar to those set forth in independent claim 1. Accordingly, the reasoning and analysis provided with respect to claim 1 apply equally to claim 13.
Regarding Claim 14 Gu-Hua
Gu-Hua discloses 14. The electronic device according to claim 13, wherein the one or more programs, when executed by the one or more processors, further cause the electronic device to (Hua, Fig. 1B, 166 “Face Enrollment Model” [0052] “The face enrollment module 166 may comprise a processor and one or more memory blocks containing, for example, instructions for execution by the processor as well as data received from another module…” [0172] “A computer readable storage medium…can contain or store a program for use by or in connection with an instruction execution system, apparatus, device, computer,…or any programmable machine or device.”), when extracting the local features:
dividing the target object into a plurality of regions based on the motion information; (Hua, [0079] “Returning to the method 400, at step 416, I.sub.i(x, y, t) is determined for each pixel in the i.sup.th image. At step 420, the i.sup.th frame from a time-stamped frame sequence undergoes segmentation into regions of interest (ROIs). Careful segmentation into various regions of interest is an important step for measuring motion interactions between face and background in terms of optical flow. This relationship is important because it characterizes a diverse set of live and spoof scenarios. [0080] “Referring briefly to FIG. 6, three primary regions of a given image are proposed for measurement and comparison:” See that Fig. 6C)
performing feature extraction on the video snippets, and integrating extracted features based on a time dimension, to obtain a plurality of time sequence convolution features; (Gu discloses performing features extraction on the video snippets, and integrating extracted features based on a time dimension, to obtain a plurality of time sequence convolution feature based, because Gu extracts inconsistency information from sampled video snippets, each comprising successive video frames, using an Intra-Snippet inconsistency information from sampled video snippets, each comprising successive video frames, using an Intra-Snippet Inconsistency Module (Intra-Sim) w. “To deal with these snip pets, an Intra-Snippet Inconsistency Module (Intra-SIM)..” “The intra-Sim performs bi-directional temporal difference operations to model intra-snippet inconsistency and aggregates the extracted temporal information. )”…generates dynamic convolutional kernels to adaptively aggregate long-range temporal information from adjacent snippets ”. Gu further generates a temporal convolution representation (Conv_U) from the temporal features (Gu, pp. 745-748, Intra-SIM discussion and Eq.(11)) and
obtaining the local features of the video snippets based on the plurality of regions and the plurality of time sequence convolution features. (Gu discloses obtaining local snippet representations (i.e., local feature) by modeling local inconsistency within each video snippet and adaptively aggregating the intra-snippet information. Gu further generates a temporal convolution (ConvU) from temporal features across adjacent video snippets. (Gu, pp. 745-746, Fig. 2 Eqs. (10) (11))
Note: The motivation set forth in the rejection of claim 1 applies equally as well to claim 14.
Regarding Claim 18 Gu-Hua
Gu-Hua discloses 18. A system comprising a device that includes:
one or more processors; (Hua, Fig. 1, 104 “processor” [0115] “one or more processors (such as the processor 104) a processor,…”) and
one or more storage apparatuses storing a video detection model and one or more programs, the one or more programs, when executed by the one or more processors, (Hua, Fig. 1B, 166 “Face Enrollment Model” [0052] “The face enrollment module 166 may comprise a processor and one or more memory blocks containing…instructions for execution by the processor…”) causing the device to:
input a global feature of a target video into the video detection model to perform category discrimination on the global feature to obtain an authenticity probability of a target object in the target video; (Gu, teaches obtaining a global feature (“global representation “) of the target video by processing the inter-snippet information using Global average pooling (GAP) (Gu, page 747). As shown in Fig. 2 (page 746), the global representation is provided to the FC layer of the video detection model to perform category discrimination, thereby determining the authenticity probability of the target video. GU further teaches determining whether the target video is Fake or Real based on the generated authenticity probability of the target video. (GU, page 746; page 750) and
obtain a detection result of the target video based on the authenticity probability. (Gu, Fig, 2, page 746 teaches determining whether the target video is Fake/Real based on the global representation generated by the network, Gu, page 750 further teaches successfully localizing fake face and spotting fake frames, thereby obtaining a detection result of the target video.)
Note: The motivation set forth in the rejection of claim 1 applies equally as well to claim 18.
Regarding Claim 20 Gu-Hua
Gu-Hua discloses 20. The system according to claim 18,
wherein the one or more programs, when executed by the one or more processors, (Hua, Fig. 1, 104 “processor” [0115] “one or more processors (such as the processor 104) a processor,…” [0172] “A computer readable storage medium…an instruction execution system, apparatus, device, computer, computing system, computer system, or any programmable machine or device.”) further cause the device to:
. . .
based on motion information of the target object in the video snippets (Hua , [0071] “One feature of the motion feature extraction module 412 is to extract motion features for each pair of time-constrained sequential frames from the time-stamped frame sequence from the video stream, and then use the frames to help characterize and quantify the motion of objects.
Hua does not expressly disclose
extract a plurality of video snippets of the target video; and
input the video snippets into the video detection model to extract local features corresponding to the video snippets, respectively, . . . , the local features being configured for representing a time sequence inconsistency of the video snippets; perform fusion on the local features to obtain the global feature of the target video.
However, Gu discloses more explicitly the following:
extract a plurality of video snippets of the target video; (Gu, page 746 “To this end, we sample the video sequence uniformly into U snip-pets, each of which contains T successive frames rather than a single frame like pervious works do.” See also Figure 2 with 3 snippets) and
input the video snippets into the video detection model to extract local features corresponding to the video snippets, respectively, (Gu, page 746 “To this end, we sample the video sequence uniformly into U snip-pets, each of which contains T successive frames rather than a single frame like pervious works do.” See also Figure 2 with 3 snippets…), . . ., the local features being configured for representing a time sequence inconsistency of the video snippets; (Gu, page 746 “Intra-Snippet Inconsistency Module(Intra-SIM) then takes frames with in each snippet to model the local inconsistency encoded in subtle motions….The long branch consists of an Intra Snippet Motion Attention(Intra-SMA) module and a path way with learnable convolutional kernels to adaptatively aggregate intra-snippet inconsistency information….To model temporal relation, Intra-SMA applies bi-directional temporal difference guided coordinate attention to make the network attend to local network.”)
perform fusion on the local features to obtain the global feature of the target video. (Gu, page 745 “ All snippets span uniformly over the entire video to form a local-to-global view. To deal with these snippets, an Intra-Snippet Inconsistency Module (Intra-SIM) and an Inter-Snippet Interaction Module (Inter-SIM) are devised and they are performed in an alternate manner. Specifically, the Intra-SIM first adopts bi-directional temporal difference operation to model the intra-snippet inconsistency…The Inter-SIM employs a new two-branch structure to establish a cross-snippet view for interaction promotion.”)
Note: The motivation set forth in the rejection of claim 1 applies equally as well to claim 20.
Claim Rejections - 35 USC § 103
Claims 3-4, 15-16, and 19 are rejected under 35 U.S.C. 103 as being unpatentable over by GU-Hua in view of Inventor waived the right to be named (CN114463805 A) (translation provided and citation given from the translated document) hereinafter “CN’805”.
Regarding Claim 3 Gu-CN’805
Gu-Hua discloses 3. The method according to claim 1,
Gu-Hua does not expressly disclose
wherein performing fusion on the local features includes:
calculating a self-attention feature among the video snippets based on the local features; normalizing the self-attention feature to obtain an initial video feature; and mapping the initial video feature through an activation function, to obtain the global feature of the target video.
However, in the same field of endeavor CN’805 discloses more explicitly the following:
wherein performing fusion on the local features (CN’805 [0043] “by fusing the local block features with the corresponding positional weight information, and performing attention enhancement processing on the fused features,..”) includes:
calculating a self-attention feature among the video snippets based on the local features; (CN’805 [0063] “ the processed feature F2 passes through a fully connected layer to obtain three features: Q, K, and V. Q is the query feature vector, K is the attribute feature vector, and V is the content feature vector. … through the softmax function, the resulting feature map represents the global connection between local blocks.”)
normalizing the self-attention feature to obtain an initial video feature; (CN’805 [0063] “The product of the query feature vector and the transposed attribute feature vector is normalized to obtain a global connection feature map between local blocks.”) and
mapping the initial video feature through an activation function, to obtain the global feature of the target video. (CN’805 [0063] “After passing through the Softmax function, the resulting feature map represents the global connection between local blocks. Multiplying this by V yields … feature V on specific blocks based on the global connection feature map.”)
Therefore, it would have been obvious to a person having ordinary skill in the art before the effective filing date of the application to modify the teachings of Gu with CN’805 to create the system of Gu as outlined above in order to “map the initial video feature through an activation function, to obtain the global feature of the target video.” as taught by CN’805. One of ordinary skill in the art would have been motivated to incorporate the Softmax activation function of CN’805 because it generates a global connection feature map among local blocks, thereby enhancing attention to forged regions and improving the accuracy of deepfake detection. CN’805 [0063]-[0065])
Regarding Claim 4 Gu-Hua-CN’805
Gu-Hua discloses 4. The method according to claim 1,
wherein determining the authenticity probability of the target object and obtaining the detection result includes:
Gu-Hua does not expressly disclose
inputting the global feature into a fully connected layer of a pre-trained video detection model to perform category discrimination on the global feature through the fully connected layer, to obtain the authenticity probability;
determining that the target video is a fake video in response to a probability that the target object is fake is greater than a preset probability threshold.
However, in the same field of endeavor CN’805 discloses more explicitly the following:
inputting the global feature into a fully connected layer of a pre-trained video detection model to perform category discrimination on the global feature through the fully connected layer, to obtain the authenticity probability; (CN’805 Claim 6 “ The third feature map is input into a pre-trained multilayer perceptron to obtain the depth forgery probability value of the face image; The depth-spoofing probability value of the face image is compared with a preset probability threshold, and the detection result of the face image is obtained based on the comparison result.”) and
determining that the target video is a fake video in response to a probability that the target object is fake is greater than a preset probability threshold. (CN’805 [0053] “step 204 can be implemented by the following method: First, the third feature map is input into a pre-trained multilayer perceptron (MLP) to obtain the deepfake probability value of the face image. Then, the deepfake probability value of the face image is compared with a preset probability threshold. Finally, the detection result of the face image is obtained based on the comparison result. That is, for face images with a deepfake probability value greater than the preset probability threshold, they are considered to have been formed by deepfake technology, thereby achieving the purpose of deepfake detection.”)
Note: The motivation set forth in the rejection of claim 3 applies equally as well to claim 4.
Regarding Claim 15 Gu-Hua-CN’805
Gu-Hua discloses 15. The electronic device according to claim 13, wherein the one or more programs, when executed by the one or more processors, further cause the electronic device to, (Hua, [0172] “A computer readable storage medium…can contain or store a program for use by or in connection with an instruction execution system, apparatus, device, computer,…or any programmable machine or device.”) when performing fusion on the local features (CN’805 [0043] “by fusing the local block features with the corresponding positional weight information, and performing attention enhancement processing on the fused features,..”): calculating a self-attention feature among the video snippets based on the local features; (CN’805 [0063] “ the processed feature F2 passes through a fully connected layer to obtain three features: Q, K, and V. Q is the query feature vector, K is the attribute feature vector, and V is the content feature vector. … through the softmax function, the resulting feature map represents the global connection between local blocks.”)
normalizing the self-attention feature to obtain an initial video feature; (CN’805 [0063] “The product of the query feature vector and the transposed attribute feature vector is normalized to obtain a global connection feature map between local blocks.”) and
mapping the initial video feature through an activation function, to obtain the global feature of the target video. (CN’805 [0063] “After passing through the Softmax function, the resulting feature map represents the global connection between local blocks. Multiplying this by V yields … feature V on specific blocks based on the global connection feature map.”)
Note: The motivation set forth in the rejection of claim 3 applies equally as well to claim 15.
Regarding Claim 16 Gu-Hua-CN’805
Gu-Hua discloses 16. The electronic device according to claim 13,
wherein the one or more programs, when executed by the one or more processors, further cause the electronic device to, (Hua, [0172] “A computer readable storage medium…can contain or store a program for use by or in connection with an instruction execution system, apparatus, device, computer,…or any programmable machine or device.”) when determining the authenticity probability of the target object and obtaining the detection result:
Gu-Hua does not expressly disclose
input the global feature into a fully connected layer of a pre-trained video detection model to perform category discrimination on the global feature through the fully connected layer, to obtain the authenticity probability; and
determine that the target video is a fake video in response to a probability that the target object is fake is greater than a preset probability threshold.
`However, in the same field of endeavor CN’805 discloses more explicitly the following
input the global feature into a fully connected layer of a pre-trained video detection model to perform category discrimination on the global feature through the fully connected layer, to obtain the authenticity probability; (CN’805 Claim 6 “ The third feature map is input into a pre-trained multilayer perceptron to obtain the depth forgery probability value of the face image; The depth-spoofing probability value of the face image is compared with a preset probability threshold, and the detection result of the face image is obtained based on the comparison result.”) and
determine that the target video is a fake video in response to a probability that the target object is fake is greater than a preset probability threshold. (CN’805 [0053] “step 204 can be implemented by the following method: First, the third feature map is input into a pre-trained multilayer perceptron (MLP) to obtain the deepfake probability value of the face image. Then, the deepfake probability value of the face image is compared with a preset probability threshold. Finally, the detection result of the face image is obtained based on the comparison result. That is, for face images with a deepfake probability value greater than the preset probability threshold, they are considered to have been formed by deepfake technology, thereby achieving the purpose of deepfake detection.”)
Note: The motivation set forth in the rejection of claim 3 applies equally as well to claim 16.
Regarding Claim 19 Gu-Hua-CN’805
Gu-Hua discloses 19. The system according to claim 18,
Gu-Hua does not expressly disclose
wherein: the video detection model includes a fully connected layer; and
the one or more programs, when executed by the one or more processors, further cause the device to, when inputting the global feature into the video detection model:
input the global feature into the fully connected layer of the video detection model for the fully connected layer to perform category discrimination based on the global feature.
However, in the same field of endeavor CN’805 discloses more explicitly the following:
wherein: the video detection model includes a fully connected layer; (CN’805 [0019] “the processing module is specifically used to input the flag features and each cascaded feature of the second feature map into a pretrained fully connected layer to obtain a query feature vector…”) and
the one or more programs, when executed by the one or more processors, further cause the device to, when inputting the global feature into the video detection model: (CN’805 [0023] “provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned deepfake detection method.”)
input the global feature into the fully connected layer of the video detection model for the fully connected layer to perform category discrimination based on the global feature. (CN’805 [0050] “….the flag features of the second feature map and each cascaded feature are input into a pre-trained fully connected layer to obtain a query feature vector, an attribute feature vector, and a content feature vector. Then, the product of the query feature vector and the transposed attribute feature vector is normalized to obtain an attention map. Finally, the attention map and the content feature vector are multiplied to obtain a third feature map. [0043] “ by fusing the local block features with the corresponding positional weight information, and performing attention enhancement processing on the fused features, global correlation features between local blocks can be extracted, and a third feature map can be obtained. [0044] 204. Classify the third feature map to obtain the detection results of the face image. [0045] Specifically, by classifying the third feature map, the detection results of the face image can be obtained. The image classification process can be achieved by using some pre-trained models or algorithms. In this embodiment, since the third feature contains global correlation features between local blocks, by classifying the global correlation features in the third feature map, the target local block features in each local block feature can be identified. Furthermore, by reverse-engineering the target local block features, the forgery features in the face image can be found, thereby achieving the purpose of deep forgery detection of face images.”)
Note: The motivation set forth in the rejection of claim 3 applies equally as well to claim 19.
Pertinent Prior Art
The prior art made of record and not relied upon is considered pertinent to applicant's
disclosure.
A. I, Jiang-qun. (CN-111932544-A)
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ASTEWAYE GETTU ZEWEDE whose telephone number is (703)756-1441. The examiner can normally be reached Mo-Fr 8:30 am to 5:30 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, William Vaughn can be reached on (571)272-3922. 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.
/ASTEWAYE GETTU ZEWEDE/Examiner, Art Unit 2481
/WILLIAM C VAUGHN JR/Supervisory Patent Examiner, Art Unit 2481
Prosecution Insights