DETAILED ACTION
Response to Amendment
Applicant’s amendments filed on 26 March 2026 have been entered. Claims 1, 9 and 13 have been amended. Claims 1-20 are still pending in this application, with claims 1, 9 and 13 being independent.
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 26 March 2026 has been entered.
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 .
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 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 of this title, 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.
Claim(s) 1-4, 8-11, 13-16 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kuniavsky et al. (US 20210286921 A1), referred herein as Kuniavsky in view of Cohen et al. (US 20240169502 A1), referred herein as Cohen, MURRAY et al. (US 20130315477 A1), referred herein as MURRAY, Bakunov et al. (US 20240297957 A1), referred herein as Bakunov and HE et al. (US 20240221215 A1), referred herein as HE.
Regarding Claim 1, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches a design system comprising: a memory storing instructions that, when executed by a processor, cause the processor to (Kuniavsky Abst: a generative design platform that generates and uses style grammars to generate product designs; FIG.6):
generate a style Kuniavsky Abst: for each of multiple products, obtaining one or more visual representations of the product and extracting, from the one or more visual representations of the product, feature values for visual features of the product.; [0004] This specification generally describes a generative design platform that generates style grammars based on features extracted from visual representations, e.g., images, of products; [0040] The generative design platform 150 can maintain a style grammar database 164, or other appropriate data structure, that includes style grammars for groups of products, e.g., brands, and/or subgroups; [0049] The style grammar generation engine 156 can generate style grammars for a product or group of products based on visual representations of the product(s)… The visual representations of the products can include images of the products and/or CAD files or CAD models of the products; [0097] The generative design platform 150 clusters the feature values for the visual features (406). The generative design platform 150 can generate clusters that represent visually similar features);
Kuniavsky does not but Cohen teaches
a style map (Cohen [0115] generate a feature vector in the form of a feature map; [0144] the cascaded modulation inpainting neural network 502 starts with an encoder E that takes the partial image and the mask as inputs to produce multi-scale feature maps from input resolution to resolution 4×4) and
a visualization model within a computation space having reduced dimensionality (Cohen [0002] a real scene, having distinct semantic areas reflecting real-world (e.g., three-dimensional) conditions; [0083] the scene-based image editing system 106 manages a two-dimensional digital image as a real scene reflecting real-world conditions; [0115] The object detection machine learning model 308 further maps each sliding window to a lower-dimensional feature).
Kuniavsky does not but MURRAY teaches
display the style map on an interface (MURRAY [0128] The user is presented with a graphic interface as shown in FIG. 4, in the form of a triangle 70 which defines the space of possible queries which are represented by all points within the triangle; [0111] The user interface allows the user to select a point only within the convex hull 70 of the three projected points. The selected (x,y) point can be projected into the (c.sub.1, c.sub.2, c.sub.3) space using Equation (9), as shown in FIG. 6. For example, a user query corresponding to the (0.4,0.4) point marked on the two dimensional style map 70 of the query space is chosen in (x,y) space); and
Kuniavsky does not but Bakunov teaches
the visualization model implements self-attention processing (Bakunov [0098] The image-grounded text encoder 704 may have a cross-attention layer inserted between a self-attention layer and a feed-forward network for each transformer block of the text encoder to inject visual information; [0099] The image encoder takes in raw pixel values of the input image and generates a lower-dimensional representation of the image that captures its visual features. The text decoder then takes this image representation as input and generates a sequence of words that describe the image in natural language. It uses causal self-attention layers instead of bi-directional self-attention layers in the text encoder. The decoder is trained using Language Modeling Loss);
Bakunov does no but HE teaches
the visualization model implements self-attention processing using a series of transformer blocks that compute attention weights per pixel and between other pixels within the image dataset to identify different styles associated with the style map (HE [0100] In step S22321, a feature vector of each pixel (i, j) in the current environmental feature map is updated based on self-attention mechanism, to obtain an updated feature vector of each pixel; [0101] for each pixel in the current environmental feature map, a feature vector of the pixel may be used as a query vector (Query), and a correlation (that is, an attention weight) between the pixel and another pixel may be obtained based on self-attention mechanism. Then, the feature vector of the pixel and a feature vector of other pixels are fused based on the correlation between the pixel and other pixels, to obtain the updated feature vector of the pixel; [0102] According to some embodiments, in step S22321, the current environmental feature map may be updated through a deformable attention (DA) mechanism. In this embodiment, for each pixel (i, j) in the current environmental feature map, the pixel is used as a reference point. Correlations (that is, attention weights) between the pixel and a plurality of neighbor pixels near the reference point are determined based on the deformable attention mechanism; [0108] the first transformer decoder includes at least one transformer layer, and each transformer layer is configured to perform one fusion on the environmental feature map and the image feature map);
Kuniavsky in view of Cohen, MURRAY, Bakunov and HE further teaches
compare images selected from the style map using scores for dimensions associated with semantic attributes to form a comparison map (Kuniavsky [0054] The design evaluation engine 154 evaluates each candidate product for one or more objectives and outputs a score for each objective based on the evaluation. One objective is conforming to the stylistic parameters of the style grammar(s) for the product design. In this evaluation, the design evaluation engine 154 can compare the visual characteristics of a candidate product design to each parameter of the style grammar; [0074] The generative design platform 150 generates a set of scores for each candidate product design (308). The scores can include a style score that represents a measure of how well the candidate product design conforms to the aesthetic characteristics of each style grammar. To determine the style score, the generative design platform 150 can compare the visual characteristics of a candidate product design to each parameter of the style grammar). The style grammar is equivalent with the semantic attributes; and the set of scores are equivalent with the comparison map.
mix visual styles of the images from the comparison map with a generative model that computes representation interpolations within a latent space (Kuniavsky [0061] FIG. 2 shows an example process 200 for generating product designs using a style grammar. In this example, the generative design platform 150 generates multiple product designs for a rim for a vehicle. In stage A, the generative design platform 150 receives data identifying a set of design parameters. The set of design inputs can be selected by a user using a client-side application 112. As described above, the inputs can include a selection of a product temple, a selection of one or more style grammars, and/or user-customizable inputs. The client-side application 112 can provide data identifying the user's selections to the generative design platform; [0064] In stage D, the generative design platform 150 runs the generative design process and automatically selects the best, e.g., highest scoring designs), the generative model outputting stylized images in an array (Cohen [0077] the server(s) 102 generates, stores, receives, and/or transmits data including digital images and modified digital images) and the visualization model and the generative model are different models (Bakunov [0151] A GAN is a generative model that comprises a generator and a discriminator; FIG.4:404; [0099] The image-grounded text decoder 706 is a type of neural network architecture that is designed to generate natural language descriptions of images; FIG. 4:412); and
communicate the array to a development system (Kuniavsky [0067] In stage G, the selected candidate designs are displayed to the user. The user interface 114 of the client-side application 112 can display a visual representation of each selected candidate rim design and optionally the score(s) used to select the candidate rim designs that are displayed to the user).
Cohen discloses systems, methods, and non-transitory computer-readable media that modify digital images via scene-based editing using image understanding facilitated by artificial intelligence. Cohen is analogous to the present patent application.
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the claimed invention to have modified Kuniavsky to incorporate the teachings of Cohen, and apply the feature map and feature of the reduced dimension into the computer-aided design (CAD) that particularly generative product design platforms.
Doing so would provide a system that facilitates flexible and intuitive editing of digital images while efficiently reducing the user interactions typically required to make such edits in systems and methods for manipulating images by comparing mapped styles and dimensions using machine learning models.
MURRAY discloses a system and method for selecting images based on photographic style by learning photographic style annotations using a data-driven approach. MURRAY is analogous to the present patent application.
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the claimed invention to have modified Kuniavsky to incorporate the teachings of MURRAY, and apply the style map interface into the computer-aided design (CAD) that particularly generative product design platforms.
Doing so would provide a more comprehensive approach to assigning aspects of photographic style to an image and which provides a user-friendly interactive method for retrieving images based on these aspects in systems and methods for manipulating images by comparing mapped styles and dimensions using machine learning models.
Bakunov discloses a system and method for automated image generation. Bakunov is analogous to the present patent application.
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the claimed invention to have modified Kuniavsky to incorporate the teachings of Bakunov and apply the image-grounded text encoder into the computer-aided design (CAD) that particularly generatives product design platforms.
Doing so would be able to fine-tuning of machine learning models based on image aspect ratios in systems and methods for manipulating images by comparing mapped styles and dimensions using machine learning models.
HE discloses a vehicle positioning method in the field of autonomous driving, deep learning and computer vision. HE is analogous to the present patent application.
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the claimed invention to have modified Bakunov to incorporate the teachings of HE and apply the deformable attention mechanism for per pixel attention weigh determining into the encoder/decoder model of Bakunov.
Doing so would provide a fused environmental feature map includes an updated feature vector of each pixel to achieve high precision images.
Regarding Claim 2, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 1, and further teaches further including instructions to compute the scores by acquiring coordinate values for the dimensions, and the semantic attributes represent interpretable qualities about the images (Kuniavsky [0027] The generative design process can include generating a set of candidate product designs for a product based on a template for the product, one or more style grammars, physical constraints in the design of the product, e.g., required materials, size, relationship between parts, etc., and/or performance objectives for the product; Cohen [0270] the scene-based image editing system 106 also determines coordinates of each object proposal relative to the dimensions of the input image 1500).
Regarding Claim 3, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 2, and further teaches wherein the instructions to compute the scores further include instructions to: estimate semantic distances of the images according to the semantic attributes by a learning model (Kuniavsky [0028] the generative design platform 150 can identify similarities between the selected candidate product designs and/or their scores and use that information in the next iteration(s). The generative design platform 150 can also identify differences between the selected candidate product designs and the non-selected product designs and use those differences as parameters in the next iteration(s); [0035] The engines can employ artificial intelligence and/or machine learning techniques to generate candidate product designs, evaluate the candidate product designs, and generate style grammars for use in generating and evaluating the product designs), wherein the visualization model, the generative model, and the learning model are the different models that form a machine learning pipeline (Bakunov [0151] A GAN is a generative model that comprises a generator and a discriminator; FIG.4:404; [0099] The image-grounded text decoder 706 is a type of neural network architecture that is designed to generate natural language descriptions of images; FIG. 4:412; [0083] The automated image generator 404 comprises a text-to-image machine learning model in the example form of a diffusion model).
Regarding Claim 4, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 3, and further teaches wherein the instructions to compute the scores further include instructions to:
classify the images to estimate the scores by a learning model (Kuniavsky [0090] In some implementations, the generative design platform 150 can use machine learning techniques to classify the perspective for each image of a product in a set of images. The generative design platform 150 can use the classifications to select the images for use in generating the style grammar for the product(s)); and
increase the dimensions to form a three-dimensional (3D) space for the images (Cohen [0297] a feature map includes a height, width, and dimension locations (H×W×D) which have D-dimensional feature vectors at each of the H×W image locations…in some instances, a feature vector is a multi-dimensional dataset that represents features depicted within a digital image. In one or more embodiments, a feature vector includes a set of numeric metrics learned by a machine learning algorithm; [0300] the scene-based image editing system 106 generates a D-dimensional image feature map f.sub.img(I)∈custom-character with a spatial size H×W extracted from a convolutional neural network-based embedding neural network).
Regarding Claim 8, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 1, and further teaches wherein the scores include factoring feedback acquired from one of decision-makers, designers, and stakeholders (Kuniavsky [0033] The user interfaces 114 also enable the user to refine or customize the generative design process; This information can also include one or more scores related to the candidate product design. The user can select candidate product designs that the user prefers and the client-send application 112 can send information identifying the selected candidate product designs to the generative design platform 150).
Regarding Claims 9-11, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches a non-transitory computer-readable medium (Kuniavsky [0122] The memory 620 stores information within the system 600. In one implementation, the memory 620 is a computer-readable medium; FIG. 6). The metes and bounds of the claims substantially correspond to the limitations set forth in claims 1-3; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Regarding Claims 13-16 and 20, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches a method (Kuniavsky Abst: a method includes…; FIG. 6). The metes and bounds of the claims substantially correspond to the limitations set forth in claims 1-5 and 8; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Claim(s) 5, 6, 12, 17 and 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kuniavsky et al. (US 20210286921 A1), referred herein as Kuniavsky in view of Cohen et al. (US 20240169502 A1), referred herein as Cohen, MURRAY et al. (US 20130315477 A1), referred herein as MURRAY, Bakunov et al. (US 20240297957 A1), referred herein as Bakunov, HE et al. (US 20240221215 A1), referred herein as HE and Cheon et al. US 20230259112 A1), referred herein as Cheon.
Regarding Claim 5, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 1, and further teaches wherein the instructions to generate the style map further include instructions to:
extract distinct styles about the images by the visualization model for the style map and the style features using the series of the transformer blocks (Kuniavsky [0094] The generative design platform 150 extracts, from the images, feature values for visual features in the images. The generative design platform 150 can use various feature extraction techniques, such as edge detection, color detection, object detection, object recognition, computer vision analysis, and/or other techniques to extract the feature values for the visual features; Bakunov [0097] The unimodal encoder 702 is configured to encode images and text separately. The unimodal encoder 702 may comprise a vision transformer as an image encoder, and a text encoder such as a transformer encoder (e.g., based on a BERT, or Bidirectional Encoder Representations from Transformers, architecture). The unimodal encoder 702 is trained using Image-Text Contrastive Loss. In contrast to negative image-text pairs, it seeks to align the feature space of the visual and text transformers by encouraging positive image-text pairs to have similar representations; [0098] The image-grounded text encoder 704 may have a cross-attention layer inserted between a self-attention layer and a feed-forward network for each transformer block of the text encoder to inject visual informatio).
The prior art does not but Cheon teaches
and identify the distinct styles and the style features using a neighbor embedding model that reduces the dimensionality (Cheon [0068] Deep learning is a class of machine learning algorithms that use a cascade of multiple layers of nonlinear processing units for feature extraction and transformation; [0191] the dimensionality reduction model may be a non-linear model, such as a t-stochastic nearest neighbor embedding model).
Cheon discloses a diagnostic tool to tool matching and comparative drill-down analysis methods for manufacturing equipment. Cheon is analogous to the present patent application.
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the claimed invention to have modified Kuniavsky to incorporate the teachings of Cheon, and apply the t-stochastic nearest neighbor embedding model for into the computer-aided design (CAD) that particularly generatives product design platforms.
Doing so would provide tool matching methods with drill-down analysis using comparative models for determining faults in manufacturing equipment in systems and methods for manipulating images by comparing mapped styles and dimensions using machine learning models.
Regarding Claim 6, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 5, and further teaches further including instructions to: alter the clustering of the images according to the distinct styles that are selected (Kuniavsky [0037] The inputs to the generative design process can include a product template that is selected by the user. The generative design platform 150 can maintain a design template database 162, or other appropriate data structure, that includes templates for multiple products, including multiple templates for each product or each type of product. The templates for a product can include variations of the product. For example, the templates for a rim for a vehicle can include templates for various size rims, templates for rims having different quantities of spokes, templates for different types of vehicles, e.g., some for sports cars and others for large trucks or heavy machinery).
Regarding Claim 12, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the non-transitory computer-readable medium of claim 9. The metes and bounds of the claim substantially correspond to the limitations set forth in claim 5; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Regarding Claims 17 and 18, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the method of claim 13. The metes and bounds of the claim substantially correspond to the limitations set forth in claims 5 and 6; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Claim(s) 7 and 19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Kuniavsky et al. (US 20210286921 A1), referred herein as Kuniavsky in view of Cohen et al. (US 20240169502 A1), referred herein as Cohen, MURRAY et al. (US 20130315477 A1), referred herein as MURRAY, Bakunov et al. (US 20240297957 A1), referred herein as Bakunov, HE et al. (US 20240221215 A1), referred herein as HE and Song et al. (US 20250022099 A1), referred herein as Song.
Regarding Claim 7, Kuniavsky in view of Cohen, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the design system of claim 1. However, Song teaches wherein the instructions to mix the visual styles further include instructions to: compute the array by the generative model partly with a latent diffusion model that is zero-shot and processes internal representations of the images into various proportions (Song [0122] CLIP can be instructed in natural language to perform a variety of classification benchmarks without directly optimizing for the benchmarks' performance, in a manner building on “zero-shot” or zero-data learning; [0160] As shown in FIG. 6, guided diffusion architecture 600 is implemented according to a pixel diffusion model. In some embodiments, guided diffusion architecture 600 is implemented according to a latent diffusion model. In a latent diffusion model, an image encoder (such as the image encoder described with reference to FIG. 5) first encodes original image 605 as image features in a latent space).
The prior art further teaches
the array having slices of the stylized images according to the various proportions that are constrained by a pose (Kuniavsky [0109] In stage A, the generative design platform 150 obtains a set of images of the car. In particular, the generative design platform 150 can obtain images from multiple perspectives from in front of the car. As shown in FIG. 5, some of the images are from angled views, some of the images are from directly in front of the car, and the height from which some of the images were captured are higher than others. By using multiple perspectives, the generative design platform 150 can extract feature values for features from the different angles at which a person may view the car).
Song discloses methods for machine learning for image generation. Song is analogous to the present patent application.
It would have been obvious for a person of ordinary skill in the art before the effective filing date of the claimed invention to have modified Kuniavsky to incorporate the teachings of Song, and apply the a latent diffusion model into the computer-aided design (CAD) that particularly generatives product design platforms.
Doing so would be able to reduce the need for expensive and large labeled datasets in systems and methods for manipulating images by comparing mapped styles and dimensions using machine learning models.
Regarding Claim 19, Kuniavsky in view of Cohen, Kuniavsky in view of Cohen, MURRAY, Bakunov and HE teaches the method of claim 13. The metes and bounds of the claim substantially correspond to the limitations set forth in claim 7; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Response to Arguments
Applicant's arguments filed on 26 March 2026, with respect to the 103 rejection have been fully considered but are moot in view of the new grounds of rejection.
Examiner notes that independent claims 1, 9 and 13 have been amended to include new limitation. Examiner finds these limitations to be unpatentable as can be found in above detail action.
On pages 10-12 of Applicant’s Remarks, the Applicant argues the dependent claims are not taught by the prior art, insomuch as they depend from claims that are not taught by the prior art. Examiner respectfully disagrees with these arguments, for the reasons discussed above.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Samantha (Yuehan) Wang whose telephone number is (571)270-5011. The examiner can normally be reached Monday-Friday, 8am-5pm.
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, King Poon can be reached on (571)272-7440. 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.
/Samantha (YUEHAN) WANG/
Primary Examiner
Art Unit 2617
Prosecution Insights