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
Claims 1-20 were rejected in the Non-Final Office action mailed on 05/28/2026. Applicant’s amended claimset, entered on 06/26/2026, amended Claims 1, 8, and 15, canceled Claims 19-20, and added new Claims 21-22. Herein this Final Office Action, Claims 1-18 and 21-22 are rejected.
Priority
Acknowledgment is made of applicant’s claim for domestic benefit to PRO #63/694,519, filed on 09/13/2024.
Response to Arguments
Applicant’s arguments filed 06/26/2026, with respect to Rejections under 35 U.S.C. 101 for Claims 1-18 and 21-22, have been fully considered and are not persuasive.
On Page 8, Applicant argues “During the interview, the Examiner indicated that amending the claims to positively recite how the claimed method improves image recognition technology-specifically by addressing the technical problem of quantifying confidence for a multi-stage image analysis pipeline where individual model uncertainties cannot be analytically combined-would help in overcoming the 35 U.S.C. § 101 rejection. This amendment clarifies that the claim is directed towards a specific, practical application that provides a technical improvement to the field of automated damage assessment and image recognition systems.” Examiner does not agree. As discussed in greater detail below, the amended claimset does not provide a patent eligible technological improvement, but merely links the abstract idea to a technology field and/or applies the abstract idea using generic computer components as a tool used in their ordinary capacity.
On Pages 8-9, Applicant argues “The amended claim is significantly more than an abstract idea because it is rooted in solving the concrete, technological problem articulated in the specification: "While individual predictions associated with components of complex systems often have well-defined confidence metrics, it may be extremely difficult to estimate confidence intervals for aggregated predictions from such systems". The newly amended claims specify a concrete technical process wherein a damage analysis pipeline includes a plurality of machine learning models that perform distinct, respective stages of image processing, feature extraction, threedimensional reconstruction, and damage detection-each producing an intermediate output with a corresponding confidence metric reflecting that model's individual uncertainty. Rather than merely aggregating or thresholding these heterogeneous uncertainties in an ad hoc manner, the claim recites a separately trained confidence metric model that learns, from prior pipeline executions and verified correctness labels, the complex, non-analytical relationship between the intermediate outputs, vehicle metadata, and the ultimate accuracy of the damage assessment. This process of training a separate machine learning model to calibrate the overall confidence of a multi-stage image recognition pipeline-where the uncertainties of its constituent models cannot be straightforwardly combined-constitutes a specific technical means that cannot be performed mentally. The features recited in the newly amended claims improve the functioning of an image recognition system itself by transforming it from an uncalibrated black box into a self evaluating system capable of automatically identifying unreliable predictions, thereby reducing false positives and improving the reliability of automated damage detection. This moves the claim firmly into the realm of a patent-eligible, computer-implemented improvement to image recognition technology.” (Emphasis added). Examiner does not agree.
MPEP 2106.05(a) states “If it is asserted that the invention improves upon conventional functioning of a computer, or upon conventional technology or technological processes, a technical explanation as to how to implement the invention should be present in the specification. That is, the disclosure must provide sufficient details such that one of ordinary skill in the art would recognize the claimed invention as providing an improvement. . . [I]f the specification explicitly sets forth an improvement but in a conclusory manner (i.e., a bare assertion of an improvement without the detail necessary to be apparent to a person of ordinary skill in the art), the examiner should not determine the claim improves technology.”
MPEP 2106.05(a)II states “However, it is important to keep in mind that an improvement in the abstract idea itself (e.g. a recited fundamental economic concept) is not an improvement in technology.”
MPEP 2106.05(a)II states “Examples that the courts have indicated may be sufficient to show an improvement in existing technology include: . . vi. Components or methods, such as measurement devices or techniques, that generate new data, Electric Power Group, LLC v. Alstom, S.A., 830 F.3d 1350, 1355, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016); . . .”
MPEP 2106.05(h) states “Examples of limitations that the courts have described as merely indicating a field of use or technological environment in which to apply a judicial exception include: . . . vi. Limiting the abstract idea of collecting information, analyzing it, and displaying certain results of the collection and analysis to data related to the electric power grid, because limiting application of the abstract idea to power-grid monitoring is simply an attempt to limit the use of the abstract idea to a particular technological environment, Electric Power Group, LLC v. Alstom S.A., 830 F.3d 1350, 1354, 119 USPQ2d 1739, 1742 (Fed. Cir. 2016); . . .”
MPEP 2106.05(f) states “When determining whether a claim simply recites a judicial exception with the words "apply it" (or an equivalent), such as mere instructions to implement an abstract idea on a computer, examiners may consider the following: (1) Whether the claim recites only the idea of a solution or outcome i.e., the claim fails to recite details of how a solution to a problem is accomplished. The recitation of claim limitations that attempt to cover any solution to an identified problem with no restriction on how the result is accomplished and no description of the mechanism for accomplishing the result, does not integrate a judicial exception into a practical application or provide significantly more because this type of recitation is equivalent to the words "apply it". See Electric Power Group, LLC v. Alstom, S.A., 830 F.3d 1350, 1356, 119 USPQ2d 1739, 1743-44 (Fed. Cir. 2016); . . . (2) Whether the claim invokes computers or other machinery merely as a tool to perform an existing process. Use of a computer or other machinery in its ordinary capacity for economic or other tasks (e.g., to receive, store, or transmit data) or simply adding a general purpose computer or computer components after the fact to an abstract idea (e.g., a fundamental economic practice or mathematical equation) does not integrate a judicial exception into a practical application or provide significantly more. . . TLI Communications provides an example of a claim invoking computers and other machinery merely as a tool to perform an existing process. The court stated that the claims describe steps of recording, administration and archiving of digital images, and found them to be directed to the abstract idea of classifying and storing digital images in an organized manner. 823 F.3d at 612, 118 USPQ2d at 1747. The court then turned to the additional elements of performing these functions using a telephone unit and a server and noted that these elements were being used in their ordinary capacity (i.e., the telephone unit is used to make calls and operate as a digital camera including compressing images and transmitting those images, and the server simply receives data, extracts classification information from the received data, and stores the digital images based on the extracted information). 823 F.3d at 612-13, 118 USPQ2d at 1747-48. In other words, the claims invoked the telephone unit and server merely as tools to execute the abstract idea. Thus, the court found that the additional elements did not add significantly more to the abstract idea because they were simply applying the abstract idea on a telephone network without any recitation of details of how to carry out the abstract idea.”
Examiner responds that the quoted portion of the Specification does not assert a non-conclusory improvement in the technology itself beyond an improvement in the recited abstract idea. As currently claimed, the combined confidence model merely provides a better way to “observe” or measure the functioning of the image processing pipeline, which is an improvement in the abstract idea itself. Secondly, the configuration of the pipeline claimed merely broadly claims the operations of machine learning models (i.e. “a plurality of machine learning models that perform respective stages of image processing, feature extraction, three-dimensional reconstruction, and damage detection”). Therefore, the machine learning models used at each stage of the pipeline, to the extent that they are additional elements (i.e. computer functionality), are merely used as a tool in their ordinary capacity. As currently claimed, and in light of the indefinite scope discussed in the rejections under 35 U.S.C. 112(b), the amended claimset fails to recite patent eligible subject matter, and therefore is rejected under 35 U.S.C. 101.
Applicant’s arguments filed 06/26/2026, with respect to Rejections under 35 U.S.C. 103 for Claims 1-18, have been fully considered and are persuasive.
Applicant’s arguments filed 06/26/2026, with respect to Rejections under 35 U.S.C. 112 for Claims 8-14, have been fully considered and are persuasive. However, Examiner notes that the amendments yield new rejections under 35 U.S.C. 112(b).
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.
Claims 1-18 and 21-22 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 applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim 1 recites the limitations of “maintaining a computing system configured to implement a damage analysis pipeline to process images of a vehicle to detect damage on the vehicle, the damage analysis pipeline including a plurality of machine learning models that perform respective stages of image processing, feature extraction, three-dimensional reconstruction, and damage detection, each machine learning model generating an intermediate output having a corresponding confidence metric that reflects an uncertainty of that machine learning model's prediction” (emphasis added) in the second paragraph, and “processing, via the computing system a plurality of intermediate outputs from machine learning models to generate a damage assessment, the intermediate outputs having corresponding confidence metrics” (emphasis added) in the third paragraph. The use of words like “pipeline,” respective “stages,” and “intermediate” output in the second paragraph, indicates that the stages of “image processing, feature extraction, three-dimensional reconstruction, and damage detection” are performed in sequence, which each output of each stage being “intermediate.” However, the third paragraph utilizes “a plurality of intermediate outputs” to generate the damage assessment.
First, the scope of “a plurality of intermediate outputs” is unclear. It is unclear as to whether the “a plurality of intermediate outputs” is intended to be a separate from the “an intermediate output” discussed in the second paragraph, or if “a plurality of intermediate outputs” is referencing the “intermediate outputs” discussed in the second paragraph. Further, if they are the same “intermediate outputs,” it is unclear as to whether the “plurality” is intended to include a subset of the intermediate outputs of each stage, or does the “plurality” reference all of the intermediate outputs.
Second, it is unclear as to whether the language of the second paragraph is intended to limit the stages to sequential operation. Looking at the second paragraph alone, a person of ordinary skill in the art would understand that the claim is limited to sequential operations of “image processing, feature extraction, three-dimensional reconstruction, and damage detection.” However, it is unclear as to whether the limitation of the third paragraph is intended to (1) claim the actual sequential “processing” of the pipeline, as the second paragraph merely “maintain[s]” the computer system configured to implement the pipeline, (2) explicitly overcome the sequential nature of the “stages” such that the “intermediate” outputs are intermediate to the generated damages assessments, or (3) the stages of the pipeline remain sequential, but additional processing of the third paragraph generates the “damage assessment.”
Finally, the second paragraph states “generating an intermediate output having a corresponding confidence metric” and the third paragraph states “the intermediate outputs having corresponding confidence metrics.” It is unclear as to whether the language of the third paragraph is intended to further limit the claim or if it is merely a reiteration of the limitation of the second paragraph.
Because the third paragraph in combination with the second paragraph creates an indefinite scope of the claim, Claim 1 is rejected under 35 U.S.C. 112(b).
Claim 1 recites the limitation ". . . automatically determininga combined confidence metric that reflects a probability that the damage assessment is accurate . . . determining that one or more combined confidence metrics are above a threshold; and presenting, on a user interface to a user of the computing system, a comprehensive damage assessment of the vehicle, the comprehensive damage assessment including identified damage having combined confidence metrics determined to be above the threshold.” (Emphasis added). Because only a singular “combined confidence metric” is introduces, the scope of “one or more combined confidence metrics” is indefinite. The “one or more combined confidence metrics” could be interpretated as (1) introducing a new feature of ““one or more combined confidence metrics” unrelated to the combined confidence metric or the plurality of “confidence metrics,” (2) intending to reference “the combined confidence metric,” but containing a typographical error, or (3) intended to reference “one or more of the confidence metrics,” but containing a typographical error. Because the scope of “one or more combined confidence metrics” is indefinite, Claim 1 is rejected under 35 U.S.C. 112(b).
Claim 1 recites “determining that one or more combined confidence metrics are above a threshold; presenting, on a user interface to a user of the computing system, a comprehensive damage assessment of the vehicle, the comprehensive damage assessment including identified damage having combined confidence metrics determined to be above the threshold” (emphasis added) in the last paragraph. It is unclear as to the scope of the “identified damage,” as there are two reasonable interpretations of the limitation. First, the “identified damage” could be defined by the combined confidence metrics above the threshold, i.e. everything above the threshold is “identified damage.” Alternatively, the “identified damage,” could be a new feature that is limited to those above a threshold, i.e. whatever the “identified damage” is, it must be of those above the threshold. Because a person of ordinary skill in the art would have multiple reasonable interpretations, the scope of this claim is indefinite. Therefore, Claim 1 is rejected under 35 U.S.C. 112(b).
Claims 2-18 and 21-22 are rejected based on dependency or recitation of similar limitations.
Claim Interpretation
Claims 1, 4, 8, 11, 15, and 18 recite a “damage analysis pipeline.” For clarity of record, the broadest reasonable interpretation of “damage analysis pipeline” is analogous to a “damage analysis process” in light of the specification (See Specification Pg. 5 Ln. 30 – Pg. 6 Ln. 14 referring to the “damage analysis pipeline” as a “multi-stage process”). No special definition of this term is asserted.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-18 and 21-22 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Step 1
Claims 1-7 and 21-22 recite a method (i.e. a process), Claims 8-14 recite a system (i.e. a machine or manufacture) Claims 15-18 recite a system (i.e. a machine or manufacture). Therefore, Claims 1-18 and 21-22 all fall within the one of the four statutory categories of invention of 35 U.S.C. 101.
Step 2A, Prong One
Independent Claim 1 recites the abstract idea of:
“. . . implement a damage analysis pipeline to process images of a vehicle to detect damage on the vehicle, the damage analysis pipeline including a plurality of . . . models that perform respective stages of image processing, feature extraction, three-dimensional reconstruction, and damage detection, each . . . model generating an intermediate output having a corresponding confidence metric that reflects an uncertainty of that . . . model's prediction;
processing, . . . a plurality of intermediate outputs from . . . models to generate a damage assessment, the intermediate outputs having corresponding confidence metrics;
automatically determining,
the combined confidence metric being determined via a confidence metric model that is a separate . . . model from the . . . models in the damage analysis pipeline,
the confidence metric model [based] on a dataset comprising, for previous damage assessments generated by the damage analysis pipeline,
(i) the intermediate outputs from the respective stages of the damage analysis pipeline and vehicle metadata associated with those damage assessments, and
(ii) verified correctness labels indicating whether each damage assessment was accurate, such that the confidence metric model [includes] a relationship between the intermediate outputs, the vehicle metadata, and the accuracy of the damage assessment to produce the combined confidence metric;
determining that one or more combined confidence metrics are above a threshold; and
presenting, . . . to a user . . . , a comprehensive damage assessment of the vehicle, the comprehensive damage assessment including identified damage having combined confidence metrics determined to be above the threshold.”
The limitations stated above are processes/ functions that under broadest reasonable interpretation covers (1) processing images of a vehicle to detect damage, (2) processing outputs from a machine learning model (i.e. the claim is limited to the “outputs” of machine learning) with corresponding confidence metrics, (3) determining a combined confidence metric, (4) presenting a comprehensive damage assessment, all of which are:
mathematical relationships (i.e. the models and creation thereof, application of a threshold, association of previous damage and other data, and relationship between variables) and mathematical calculations (i.e. outputs of models, probability of accuracy, and application of a threshold), which are mathematical concepts, an abstract idea, under MPEP 2106.04(a)(2)I,
managing personal behavior by following rules and interacting between people (i.e. receiving and displaying information and coordinating and directing a service) and commercial or legal interactions (i.e. damage assessment and accuracy thereof is at least a part of “marketing or sales activities or behaviors”), which are certain methods of organizing human activity, an abstract idea, under MPEP 2106.04(a)(2)II, and
observations (i.e. processing an image of a vehicle), evaluation (i.e. damage analysis pipeline, feature extraction, three-dimensional reconstruction, damage detection, and confidence metric), judgment (i.e. application of threshold), opinion (i.e. presenting damage assessment), which are mental processes, an abstract idea, under MPEP 2106.04(a)(2)III.
The mere the recitation of additional elements (i.e., generic computer components implementing the identified abstract idea) does not take the claim out of the mathematical concepts, certain methods of organizing human activity, and mental processes groupings. MPEP 2106.04(d). Therefore, Claim 1 recites an abstract idea.
Step 2A, Prong Two
The judicial exception is not integrated into a practical application. Claim 1 as a whole amounts to: (i) merely invoking generic components as a tool to perform the abstract idea or “apply it” (or an equivalent) and (ii) generally links the use of a judicial exception to a particular technological environment or field of use. The claim recites the additional elements of:
(i) “computer system,”
(ii.a) “machine learning” including (ii.b) “train[ing]” and (ii.c) “learn[ing],” and
(iii) “user interface.”
The additional elements of (i) computer system (Fig. 21 and pg. 44 ln. 22-30 shows that computer system 2100 can be a single device, such as a smart phone, or multiple devices such as a remote camera system and a desktop PC.), (ii) machine learning (pg. 6 and ln. 01-04 “In executing such a multi-stage process, the computing system may employ computer vision algorithms, Convolutional Neural Networks (CNNs) or other deep learning architectures at a variety of these stages.” Pg. 6 ln. 22-25 shows “The architecture of the combined confidence metric generation model (e.g. combined confidence metric generation model 200 of Figure 2), may vary across implementations. For example, the combined confidence metric generation model may be a neural network, combination of neural networks, or other deep learning architecture.”), and (iii) user interface (Fig. 21 and pg. 43 ln. 31 - pg. 44 ln. 01 shows “The interface 2111 may include separate input and output interfaces, or may be a unified interface supporting both operations. Examples of input and output interfaces can include displays, audio devices, cameras, touch screens, buttons and microphones.”), are recited at a high-level of generality, such that, when viewed as whole/ordered combination (Fig. 21, pg. 6 ln. 01-25, and pg. 43 ln. 31 – pg. 44 ln. 30 shows elements in combination, e.g. smart phone), they amount to no more than mere instruction to apply the judicial exception using generic computer components or “apply it” (See MPEP 2106.05(f)). Additionally, when viewed with the abstract idea in the claim as a whole, the additional elements do not provide a patent eligible improvement to technology per MPEP 2106.05(a).
The (i) computer system, (ii) machine learning, and (iii) user interface, when viewed as whole/ordered combination (Fig. 21, pg. 6 ln. 01-25, and pg. 43 ln. 31 – pg. 44 ln. 30 shows elements in combination, e.g. smart phone.), does no more than generally link the use of the judicial exception to a particular technological environment or field of use (i.e. computer environment) (See MPEP 2106.05(h)).
Accordingly, these additional elements, when viewed as a whole/ordered combination (Fig. 21, pg. 6 ln. 01-25, and pg. 43 ln. 31 – pg. 44 ln. 30 shows elements in combination, e.g. smart phone.), do not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea. Thus, the claim is directed to an abstract idea.
Step 2B
As discussed above with respect to Step 2A Prong Two, the additional elements amount to no more than: (i) “apply it” (or an equivalent) and (ii) generally link the use of a judicial exception to a particular technological environment or field of use, and are not a practical application of the abstract idea. The same analysis applies here in Step 2B, i.e., (i) merely invoking the generic components as a tool to perform the abstract idea or “apply it” (See MPEP 2106.05(f)) and (ii) generally linking the use of a judicial exception to a particular technological environment or field of use (See MPEP 2106.05(h)), does not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B.
Therefore, the additional elements of the (i) computer system, (ii) machine learning, and (iii) user interface, do not integrate the abstract idea into a practical application at Step 2A or provide an inventive concept at Step 2B. Thus, even when viewed as a whole/ordered combination (Fig. 21, pg. 6 ln. 01-25, and pg. 43 ln. 31 – pg. 44 ln. 30 shows elements in combination, e.g. smart phone), nothing in the claims adds significantly more (i.e., an inventive concept) to the abstract idea. Thus, the claim is ineligible.
Dependent Claims 2-7 recite the abstract idea of:
“ . . . correcting one or more of the outputs associated with the combined confidence metrics that are below the threshold; and presenting the corrected outputs . . .” (Claims 2).
“. . . wherein determining the combined confidence metric is further based on prior information.” (Claims 3).
“. . . wherein determining the combined confidence metric is further based on final outputs of the damage assessment pipeline and data associated with the vehicle.” (Claims 4).
“. . . wherein the data associated with the vehicle includes mileage of the vehicle, age of the vehicle, make of the vehicle, and/or model of the vehicle.” (Claims 5).
“. . . wherein the final outputs include damage location, damage severity, and/or damage type.” (Claims 6).
“. . . wherein the threshold is dynamically adjustable by users . . .” (Claims 7).
Dependent Claims 2-7, have been given the full two-prong analysis including analyzing the further elements and limitations, both individually and in combination. When analyzed individually and in combination, these claims are also held to be patent ineligible under 35 U.S.C. 101. The further limitation of Claims 2-7 fail to establish claims that are not directed to an abstract idea because the further limitations include (1) correcting outputs and presenting corrected outputs, (2) using certain information for the combined confidence metric, (3) the final output including certain information, and (4) adjusting the threshold by a user, which are a part of the abstract idea. The further elements of Claims 2-7 (i.e. “user interface of a display device” Claim 2 and “computing system” of Claim 7) fails to establish claims that are not directed to an abstract idea because the elements merely recite additional generic computer components similar to the generic computer components of Claim 1 and generally link the abstract idea to a particular technology or field of use (i.e. computer environment) just as in Claim 1. The organization of the further limitations of Claims 2-7 fail to integrate an abstract idea into a practical application just as discussed above for Claim 1. Additionally, performing the abstract idea of Claim 1 as recited in each of the further limitations of Claims 2-7, individually or in combination, does not (1) impose any meaningful limits on practicing the abstract ideas, or (2) provide improvements to the functioning of computing systems or to another technology or technical field, just as discussed above regarding Claim 1. Therefore, Claims 2-7 amount to mere instructions to implement the abstract idea (1) using generic computer components—using the computer, in its ordinary capacity, as a tool to perform the abstract idea, and (2) generally linked to a particular technology or field of use. Because the claims merely use a computer, in its ordinary capacity in a particular field of use, as a tool to perform the abstract idea cannot provide an inventive concept, the elements and limitations of Claims 2-7 fail to establish that the claims provide an inventive concept, just as in Claim 1. Therefore, Claims 2-7 fails the Subject Matter Eligibility Test and are consequently rejected under 35 U.S.C. 101.
Claims 8-14 recite elements and limitations that are substantially similar to Claims 1-7. Claims 1-7 recite a method embodied by the elements and limitations of Claims 8-14. Therefore, Claims 8-14 are rejected under 35 U.S.C. 101 just as Claims 1-7 are rejected under 35 U.S.C. 101 as discussed above.
Claims 15-20 recite elements and limitations that are substantially similar to Claims 1-6. Claims 1-6 recite a method embodied by the elements and limitations of Claims 15-20. Therefore, Claims 15-20 are rejected under 35 U.S.C. 101 just as Claims 1-6 are rejected under 35 U.S.C. 101 as discussed above.
Dependent Claims 21-22 recite the abstract idea of:
“ . . . wherein automatically determining the combined confidence metric comprises:
preprocessing inputs to the confidence metric model, the inputs including the intermediate outputs and the vehicle metadata, wherein
categorical values among the inputs are preprocessed using [mathematical transformations] and
numerical values among the inputs are preprocessed to have a designated mean and variance;
generating one or more synthetic inputs for the confidence metric model by computing a cross product of at least two categorical inputs; and
predicting the combined confidence metric using the preprocessed inputs and the synthetic inputs.” (Claim 21).
“. . . automatically detecting that a component of the damage analysis pipeline has been modified;
in response to automatically detecting the modification, re-executing the modified damage analysis pipeline on a set of vehicle images to generate new damage assessment results;
receiving new verified correctness labels for the new damage assessment results; and
re-[creating] the confidence metric model using the new damage assessment results and the new verified correctness labels.” (Claim 22).
Dependent Claims 21-22, have been given the full two-prong analysis including analyzing the further elements and limitations, both individually and in combination. When analyzed individually and in combination, these claims are also held to be patent ineligible under 35 U.S.C. 101. The further limitation of Claims 21-22 fail to establish claims that are not directed to an abstract idea because the further limitations include (1) preprocessing inputs to the confidence model including certain types of information (Claim 21), (2) transforming the categorical values into a binary matrix (Claim 21), (3) numerical values have a designated mean and variance (Claim 21), (4) generating synthetic inputs by computing a cross product (Claim 21), (5) predicting combined confidence metric using preprocessed inputs and synthetic inputs (Claim 21), (6) detecting that the pipeline has been modified (Claim 22), (7) in response to observing a modification, re-executing the modified pipeline on a set of images to generate new results (Claim 22), (8) receiving new labels of the new results (Claim 22), and (9) re-creating the confidence metric model using the new labels and results (Claim 22), which are a part of the abstract idea. The further elements of Claims 21-22 (i.e. “one-hot encoding” of Claim 21 and “re-training” of Claim 22) fails to establish claims that are not directed to an abstract idea because the elements merely recite additional generic computer components similar to the generic computer components of Claim 1 and generally link the abstract idea to a particular technology or field of use (i.e. computer machine learning environment) just as in Claim 1. The organization of the further limitations of Claims 21-22 fail to integrate an abstract idea into a practical application just as discussed above for Claim 1. Additionally, performing the abstract idea of Claim 1 as recited in each of the further limitations of Claims 21-22, individually or in combination, does not (1) impose any meaningful limits on practicing the abstract ideas, or (2) provide improvements to the functioning of computing systems or to another technology or technical field, just as discussed above regarding Claim 1. Therefore, Claims 21-22 amount to mere instructions to implement the abstract idea (1) using generic computer components—using the computer, in its ordinary capacity, as a tool to perform the abstract idea, and (2) generally linked to a particular technology or field of use. Because the claims merely use a computer, in its ordinary capacity in a particular field of use, as a tool to perform the abstract idea cannot provide an inventive concept, the elements and limitations of Claims 21-22 fail to establish that the claims provide an inventive concept, just as in Claim 1. Therefore, Claims 21-22 fails the Subject Matter Eligibility Test and are consequently rejected under 35 U.S.C. 101.
Reasons for No Art Rejection
Claims 1-18 and 21-22 are not rejected over the prior art of record.
The Closest prior art of record is:
US-2022129860-A1 (“Abhilash”);
US-2023069070-A1 (“Ranca”);
US-20190073641-A1 (“Utke”);
US-20180260793-A1 (“Li-793”);
US-20210263525-A1 (“Das”);
“Ensemble Deep Learning for Automated Damage Detection of Trailers at Intermodal Terminals” (“Cimili” 01/31/2024 https://www.mdpi.com/2071-1050/16/3/1218);
US-20220358775-A1 (“Hantehzadeh”);
US-20220019858-A1 (“Hyland”);
US-20250190952-A1 (“Byers”);
US-20250054124-A1 (“Olson”);
US-20200111061-A1 (“Stucki”);
US-20190095764-A1 (“Li-764”); and
WO-2021167998-A1 (“Huts”).
The Following is an examiner’s statement of reasons for no art rejection:
Abhilash discloses image processing system that determines vehicle damage information. The system includes separate pre-processing, part detection, and damage detection stages that utilize a convolution neural network or other AI/ML model trained with labeled datasets. The system applies an accuracy value (e.g. 75%), and aggregates the inferences to create a final inference. However, the individual inferences are of frames of a video, which are then filtered based on the accuracy values. Therefore, although Abhilash does teach the concept of multiple machine learning models used to detect vehicle damage, and combining intermediate confidence values to create an overall confidence value, Abhilash does not teach the specific configuration of the machine learning models claimed.
Ranca discloses multiple trained machine learning models used as part of a computer vision process to provide repair estimates based on received photos of damage. The computer vision process applies the combined confidence value of the damage vector to a threshold to determine if more information is needed. The damage vector is determined by first applying a part classifier machine learning model to the received photo, then applying multiple repair machine learning models for each type of repair (e.g. part replacement, paint, labor, ect.). Therefore, although Ranca does teach the concept of multiple machine learning models used to detect vehicle damage, and combining intermediate confidence values to create an overall confidence value, Ranca does not teach the specific configuration of the machine learning models claimed.
Utke discloses image analysis, including a confidence variable, to assess vehicle damage. The damage detection application includes object recognition model, damage model, angle model, image selection model, parts cluster damage model, and repair/replace model. However, Utke merely applies a confidence score to each frame of the input video, which does not teach the claimed intermediate confidence metrics, or the combination thereof with the trained confidence model.
Li-793 discloses the use of a confidence score throughout a model used to assess vehicle damage. The image processing includes sequential steps of: image cleaning to remove background and reflection, image alignment to align the image to a reference image, image segmentation to divide the vehicle into vehicle parts, and damage assessment based on (a) edge distribution, (b) texture, and (c) spatial correlation. Each of these stages can be performed with machine learning models trained with a tagged/annotated dataset. Thus, Li-793 teaches the multiple stages of machine learning models claimed (i.e. image processing, feature extraction, three-dimensional reconstruction, and damage detection). Li-793 further discloses the use of a confidence score heatmap, developed by a machine learning model. However, Li-793 does not explicitly teach that (1) the intermediate outputs of each of the damage assessment machine learning models includes including a confidence score, or (2) the training of the confidence machine learning model with the previous damage assessments, as claimed. More specifically, Li-793 does not teach the combined confidence metric as claimed.
Das discloses multiple image recognition pipelines, which are combined into a fusion pipeline. Das further discloses a combined confidence model. However, Das is directed to image recognition on board a vehicle, and does teach damage detection of a vehicle. Additionally, the fusion pipeline of Das combines image, radar, and lidar to input into the combined model.
Cimili discloses use of a confidence matrix in image analysis of vehicle damage. However, the confidence matrix does not use intermediate outputs of multiple machine learning models in a multi-stage damage analysis pipeline.
Hantehzadeh discloses image processing systems for detecting vehicle segments from vehicle images. The process includes applying a set of trained models to an input image that could operate in sequential order. The process includes an identifier (year/make/model) model, tagging model that tags the view shown in the image (e.g. front view), heat mapping model that determines the probability of damage at each pixel, damage typing model that detects the probabilities of multiple types of damage, and segmenting model that determines the boarder of an object (e.g. right front headlight), along with a warper for warped images and stitcher for stitched images used by the damage detailer that measures damage size and location. Hantehzadeh teaches that each of the models can output probabilities or confidence values. Therefore, Hantehzadeh teaches the multi-stage machine learning models with intermediate outputs including a confidence metric. However, Hantehzadeh does not explicitly teach a separate combined confidence machine learning model as claimed.
Hyland discloses a vehicle damage assessment process that inputs an image into a trained machine learning model. The model predicts the probability (i.e. confidence metric) of each repair operation that could be performed on a vehicle, then applies a threshold to determine what vehicle repair operation is needed. This prediction can be reviewed by a human to retrained the model based on whether the human agrees or disagrees with the prediction. Although Hyland teaches a plurality of confidence metrics, and training a confidence model with previous determinations, Hyland does not teach the multiple machine learning models that output intermediate values including a confidence metric.
Byers discloses using multiple machine learning models to evaluate and estimate vehicle damage and a confidence level. However, the damage analyzer is multi-model, meaning that one machine learning model determines the damage based on image input, and the other machine learning models determine damage based on other modes of input, such as vehicle telematics or verbal/written description of the damage.
Olson discloses a repair prediction apparatus comprising separate modules including an artificial intelligence (AI) model, a zone module, a confidence score module, an image analysis module, an appointment module, and a GUI module. The confidence score module determines a confidence score based on the quality of the submitted image, the damage type, the location of the damage, data besides the image input by the user, comparison to other images, or any combination thereof. Although the confidence score module is separate from the other modules, and can combine multiple inputs, Olson teaches the AI model as separate from the modules. Olson teaches the AI model as recognizing the damage from images. Therefore, Olson does not teach the claimed multi-stage pipeline of machine learning models with intermediate confidence outputs that are input into a separate combined confidence machine learning model.
Stucki discloses a visual intelligence engine that receives images of damaged vehicles and outputs repair steps and cost estimation. The visual intelligence engine includes an aggregation engine that receives inputs from the other sub-engines to generate a list of damaged parts. The other sub-engines include an image categorization engine that provides macro-level categorization of the images (e.g. odometer image, VIN image, interior image, full-view vehicle image, close-up door image, ect.), side detection engine that identifies the perspective of the image (i.e. taken from the left, right, front, or back side of the vehicle), object detection engine that utilizes instance segmentation to identify the parts and damage of the vehicle (e.g. door, hood, fender, ect), model detection engine that identifies the make and model of the vehicle, damage attribute engine that determines which of the parts are damaged, and claim-level classification engine to identify damaged components. Thus, Stucki teaches a multi-stage damage assessment pipeline. Further, the aggregation engine uses a machine learning model trained on historical claim data, and the aggregation engine provides a combined confidence score. However, Stucki does not teach a specific way that the aggregation engine combines different confidence levels, but merely that the aggregation engine reconciles differing determinations of each sub-engine.
Li-764 discloses using machine learning models to detect damage to property based on images input to the models. The system includes multiple feature extraction models, multiple classification models, and a metal model. The feature extraction models extracts geometric properties (e.g. lines and shapes) of the objects in the images. The classification models distinguishes between damaged and undamaged regions of the extracted features. The meta model is trained separately from the other models, and outputs a confidence score from each of the classification model outputs, and then combines them to aggregate the classification model outputs. However, Li-764 has three major distinctions from the instant claims. First, the multiplicity of feature extraction and classification models do not teach the pipeline comprising image processing, feature extraction, three-dimensional reconstruction, and damage detection. Second, the individual confidence values are not a part of the intermediate outputs of what would teach the pipeline models. Third, although Li-764 mentions that the damaged property could be a vehicle, the meta model of Li-764 does not teach the vehicle-centric data (i.e. vehicle metadata) used to train what would teach the combined confidence model.
Huts discloses a combination of data analytics and computer vision. The system utilizes a performance (e.g. accuracy) machine learning model to measure the accuracy of a plurality of image analysis machine learning models. Huts further teaches an application of vehicle damage detection among other. However, Huts does not teach the vehicle-centric data (i.e. vehicle metadata) used to train what would teach the combined confidence model.
Generally, the closest prior art teaches either (1) use of multiple machine learning models and confidence metric for damage detection without specific configuration of claimed pipeline (Abhilash, Ranca, and Li-764), (2) use of confidence metric without specific configuration of claimed pipeline (Abhilash, Ranca, Das, Cimili, Hyland, Byers, Olson, and Huts), or (3) use of specific configuration of claimed pipeline without claimed confidence limitations (Utke, Li-793, Hantehzadeh, and Stucki).
With respect to independent Claims 1, 8, and 15, the closest prior art, taken individually and in an ordered combination, does not explicitly or implicitly disclose the specific ordered combination of elements claimed. Although the scope of the function of each of the pipeline machine learning models, as they are only limited by the function of the stage (i.e. image processing, feature extraction, three-dimensional reconstruction, and damage detection), the claim explicitly includes a damage analysis pipeline including four machine learning models at each stage. The claims further includes a separate confidence metric model that outputs a combined confidence metric. Thus, the claims includes five machine learning models, i.e. four as part of the damage analysis pipeline and the fifth trained on previous damage assessments of the pipeline and other vehicle damage related data, to determine a confidence metric. It would not be obvious to combine the prior art of record to produce the five machine learning models as specifically claimed. Therefore, independent Claims 1, 8, and 15 are not rejected by the prior art of record.
Dependent Claims 2-7 and 21-22 depend on Claim 1, dependent Claims 9-14 depend on Claim 8, and dependent Claims 16-18 depend on Claim 15, and therefore are also not rejected via dependency.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEW PARKER GOODMAN whose telephone number is (571) 272-5698. The examiner can normally be reached on Monday-Thursday from 9:30 AM ET to 6:00 PM ET. If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Jeffrey Zimmerman, can be reached at telephone number (571) 272-4602. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://portal.uspto.gov/external/portal. Should you have questions about access to the Private PAIR
system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free).
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.
/MATTHEW PARKER GOODMAN/Examiner, Art Unit 3628
/DANIEL VETTER/Primary Examiner, Art Unit 3628
