DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
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and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114.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 02/23/2026 has been entered.
Allowable Subject Matter
Claims 21-22 are objected to as being dependent 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.
The following is a statement of reasons for the indication of allowable subject matter: The present applicant has been thoroughly reviewed. The following claim language in combination with the limitations of the independent claims would place the Application in condition for allowance:
wherein the other types of attributes comprise color-irrelevant attributes, and wherein training the image-text embedding model comprises: training, by the computing system, the image-text embedding model using the modified image, the text prompt, and the negative training examples to generate disentangled embeddings that separate color attributes of the object from the color-irrelevant attributes COMBINED WITH wherein the color-irrelevant attributes comprise object geometry and texture, and wherein the disentangled embeddings allow for color-independent object recognition.
Response to Arguments
Applicant's arguments filed 02/23/2026 have been fully considered but they are not persuasive.
The Applicant alleges the following on page 9 of the remarks: “Under MPEP § 2143.03, a proper rejection requires a clear articulated reasoning for combining teachings. As established in KSR Int'l Co. v. Teleflex Inc., and reiterated in MPEP § 2141, the Examiner must provide a "reasoned explanation" for why a person of ordinary skill in the art (PHOSITA) would have been motivated to modify a specific primary reference with the teachings of a secondary reference. The Office Action does not explain why a PHOSITA would start with Aggarwal and modify it with Udhayanan. Instead, the Office Action jumps between the two documents as teaching the above-mentioned recitations only after having read Applicant's disclosure. This "piecemeal" assembly lacks a logical framework and fails to consider whether the features of Udhayanan are even compatible with the structural environment of Aggarwal. Because the Office Action relies on a piecemeal assembly of references without a clear primary anchor or a non-hindsight-based motivation, the rejection under 35 U.S.C. 103 is improper and should be withdrawn.” The examiner is not persuaded. The Applicant' s response is merely a citation of case law and fails to provide any explanation of how it is relevant to the examiner' s rejection. MPEP 710.01 discusses applicant responses that fail to point out the examiner' s supposed errors and therefore do not comply with 37 CFR 1.111(b). Moreover, the examiner specifically notes that the references, Aggarwal, Udhayanan and Dammu, teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image processing. This close relation between the references highly suggests an expectation of success. Accordingly, the examiner maintains the rejection.
The Applicant alleges the following on pages 9-10 of the remarks: “Applicant respectfully submits that the combination of Aggarwal and Udhayanan fails to teach or suggest the specific "training, by the computing system, an image-text embedding model comprising an image encoder and a text encoder using the modified image, the text prompt, and negative training examples to generate disentangled embeddings that separate color attributes of the object from other types of attributes," as recited in amended claim 1”. The examiner is not persuaded. The combination of Aggarwal, Udhayanan and Dammu specifically teaches the Applicant’s claim language. Moreover, the combination of Aggarwal, Udhayanan and Dammu specifically teaches the following:
training, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023),
an image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065)
comprising an image encoder (See Aggarwal Paragraph 0031)
and a text encoder (See Aggarwal Paragraph 0031)
using the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041),
the text prompt (See Udhayanan Paragraphs 0034; 0064-0065),
and negative training examples (See Aggarwal Paragraph 0066)
to generate disentangled embeddings (See Dammu Paragraph 0040l 0051) that separate color attributes of the object from other types of attributes
(See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041).
MPEP § 2106 states Office personnel are to give claims their broadest reasonable interpretation in light of the supporting disclosure. In re Morris, 127 F.3d 1048, 1054-55, 44 USPQ2d 1023, 1027-28 (Fed Cir. 1997). Accordingly, the examiner maintains the rejection.
Claim Rejections - 35 USC § 103
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.
Claim(s) 1-3, 7-8, 11, 14-18 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Aggarwal US 20220343561 in view of Udhayanan, US20250022263 and in further view of Dammu, US 20220269893.
Claim 1:
Aggarwal discloses a computer-implemented method (See Aggarwal Abstract1). However, Aggarwal failed to explicitly disclose the following:
obtaining, by the computing system, a text prompt that describes the modified image, wherein the text prompt includes one or more text tokens;
and training, by the computing system, an image-text embedding model using the modified image and the text prompt.
Udhayanan discloses the following features:
a text prompt (See Udhayanan Paragraphs 0033-0036; 0064--00652)
one or more text tokens (See Udhayanan Paragraphs 0033-0036; 0043-0044);
an image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065).
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have further modified Aggarwal’s system/ method of image editing by incorporating text indicating a desired change to be made on the image, as taught by Udhayanan, thereby modifying images quickly and more efficiently. Additionally, the references, Aggarwal and Udhayanan, teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image processing. This close relation between the references highly suggests an expectation of success.
Additionally, Aggarwal and Udhayanan failed to disclose disentangled embeddings, however, Dammu discloses this feature in Paragraph 0040; 0051. It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have further modified Aggarwal’s and Udhayanan system/ method, to incorporate disentangling embeddings, thereby generating face images, more efficiently. Additionally, the references, Aggarwal, Udhayanan, and Dammu teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image (face) processing. This close relation between the references highly suggests an expectation of success.
As modified:
The combination of Aggarwal, Udhayanan and Dammu discloses the following:
a computer-implemented method to improve a robustness of an image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) to color specificity (See Aggarwal Abstract), the method comprising:
obtaining, by a computing system comprising one or more computing devices (See Aggarwal Figure 1, Item 105; Paragraph 0023), an initial image (See Aggarwal Figure 2, Item 200; Paragraphs 0035-0041) that depicts an object having a first color (See Aggarwal Figure 2, Item 205; Paragraphs 0035-0041);
modifying, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), color values of the initial image to generate a modified image in which the object has a second color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041);
obtaining, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), a text prompt (See Udhayanan Paragraphs 0034; 0064-00653) that describes the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041), wherein the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) includes one or more text tokens (See Udhayanan Paragraphs 0033-0036; 0043-0044) that correspond to the second color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041);
training, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), an image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) comprising an image encoder (See Aggarwal Paragraph 0031) and a text encoder (See Aggarwal Paragraph 0031) using the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041), the text prompt (See Udhayanan Paragraphs 0034; 0064-0065), and negative training examples (See Aggarwal Paragraph 0066) to generate disentangled embeddings (See Dammu Paragraph 0040l 0051) that separate color attributes of the object from other types of attributes (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041);
and storing, by the computing system, the image-text embedding model in memory (See Udhayanan Paragraphs 0034; 0064-0065).
Claim 2:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein obtaining the initial image (See Aggarwal Figure 2, Item 200; Paragraphs 0035-0041) comprises: obtaining, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), an initial prompt, wherein the initial prompt includes one or more text tokens (See Udhayanan Paragraphs 0035-0036) that correspond to the first color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041); and processing, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), the initial prompt (See Udhayanan Paragraphs 0035-0036) with a text-to-image generation model (See Udhayanan Figure 2; Paragraphs 0025-0026; 0049-0051) to generate the initial image(See Aggarwal Figure 2, Item 200-220; Paragraphs 0035-0041).
Claim 3:
The combination of Aggarwal, Udhayanan and Dammu discloses modifying (See Aggarwal Figure 2, Item 220), by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), the initial prompt to generate the text prompt (See Udhayanan Paragraphs 0035-0036), wherein said modifying (See Aggarwal Figure 2, Item 220) comprises replacing the one or more text tokens (See Udhayanan Paragraphs 0035-0036) that correspond to the first color (See Aggarwal Figure 2, Item 205; Paragraphs 0035-0041) with the one or more text tokens (See Udhayanan Paragraphs 0035-0036) that correspond to the second color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041).
Claim 7:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the second color comprises a brand-specific color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041).
Claim 8:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the one or more text tokens (See Udhayanan Paragraphs 0035-0036) that correspond to the second color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) comprise rare-text tokens (See Udhayanan Paragraphs 0035-0036).
Claim 9:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the image encoder (See Aggarwal Paragraph 0031) configured to process the modified image to generate an image embedding (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) and the text encoder (See Aggarwal Paragraph 0031) is configured to process the text prompt to generate a text embedding (See Udhayanan Paragraphs 0033-0036; 0064-0065).
Claim 11:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein training, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) using the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) and the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) comprises: generating, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), one or more hard negative images (See Aggarwal Paragraph 0066) that depict the object having one or more third colors (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041); and applying, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), a hard negative loss (See Aggarwal Paragraph 0066) between the image embedding (See Aggarwal Figure 2, Item 220; Paragraphs 0020; 0035-0041) generated by the image encoder (See Aggarwal Paragraph 0031) for the modified image and one or more image embeddings (See Aggarwal Figure 2, Item 220; Paragraphs 0020; 0035-0041) generated by the image encoder (See Aggarwal Paragraph 0031) for the one or more hard negative images (See Aggarwal Paragraph 0066).
Claim 14:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the second color (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) is selected by a user from an RGB color space (See Aggarwal Paragraph 0019).
Claim 15:
Claim 15 is rejected on the same basis as claim 1.
Claim 16:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the computer system (See Aggarwal Figure 1, Item 105; Paragraph 0023) is configured to use the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) to perform text-to-image retrieval or text-to-image generation (See Udhayanan Figure 2; Paragraphs 0025-0026; 0049-0051).
Claim 17:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the computer system (See Aggarwal Figure 1, Item 105; Paragraph 0023) is configured to use the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) to perform image-to-text retrieval (See Udhayanan Figure 2; Paragraphs 0025-0026; 0049-0051).
Claim 18:
The combination of Aggarwal, Udhayanan and Dammu discloses wherein the computer system (See Aggarwal Figure 1, Item 105; Paragraph 0023) is configured to use the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) to perform image-to-image retrieval (See Udhayanan Figure 2; Paragraphs 0025-0026; 0049-0051).
Claim 20:
Claim 20 is rejected on the same basis as claim 1.
Claim(s) 4, 5 and 10 are rejected under 35 U.S.C. 103 as being unpatentable over Aggarwal US 20220343561 in view of Udhayanan, US20250022263 in view of Dammu, US 20220269893 and in further view of Park, US 20240281924.
Claim 4:
The combination of Aggarwal, Udhayanan and Dammu failed to disclose a pre-trained language model. However, Park discloses the feature in Paragraph 0069. It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have further modified Aggarwal’s system/method of image editing by incorporating text indicating a desired change to be made on the image, as taught by Udhayanan with the pre-trained language model, as taught by Dammu’s deep learning models, and as taught by Park, thereby modifying images, more efficiently (See Park’s Summary of Invention). Additionally, the references, Aggarwal, Udhayanan, Dammu and Park, teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image processing. This close relation between the references highly suggests an expectation of success.
As modified:
The combination of Aggarwal, Udhayanan, Dammu and Park discloses wherein obtaining the initial prompt (See Udhayanan Paragraphs 0034; 0064-0065) comprises generating the initial prompt (See Udhayanan Paragraphs 0034; 0064-0065) using a pre-trained language model (See Park Paragraph 0069).
Claim 5:
The combination of Aggarwal, Udhayanan and Dammu failed to disclose a denoising diffusion model. However, Park discloses the feature in Paragraph 0182. It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have further modified Aggarwal’s system/method of image editing by incorporating text indicating a desired change to be made on the image, as taught by Udhayanan with the pre-trained language model, as taught by Dammu’s deep learning models, and as taught by Park, thereby modifying images, more efficiently (See Park’s Summary of Invention). Additionally, the references, Aggarwal, Udhayanan and Dammu, teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image processing. This close relation between the references highly suggests an expectation of success.
As modified:
The combination of Aggarwal, Udhayanan, Dammu and Park discloses wherein the text-to-image generation model (See Udhayanan Paragraphs 0033-0036; 0064-0065) comprises a denoising diffusion model (See Park Paragraph 0182).
Claim 10:
The combination of Aggarwal, Udhayanan and Dammu failed to disclose a CLIP loss. However, Park discloses the feature in Paragraph 0187. It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have further modified Aggarwal’s system/method of image editing by incorporating text indicating a desired change to be made on the image, as taught by Udhayanan with the pre-trained language model, as taught by Dammu’s deep learning models, and as taught by Park, thereby modifying images, more efficiently (See Park’s Summary of Invention). Additionally, the references, Aggarwal, Udhayanan, and Dammu teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image processing. This close relation between the references highly suggests an expectation of success.
As modified:
The combination of Aggarwal, Udhayanan, Dammu and Park discloses wherein training, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) using the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) and the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) comprises applying, by the computing system a CLIP loss (See Park Paragraph 0187) between the image embedding and the text embedding (See Park Paragraph 0047).
Claim(s) 12 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Aggarwal US 20220343561 in view of Udhayanan, US20250022263 in view of Dammu, US 20220269893 in view of Park, US 20240281924 and in further view of Zhao, CN116258652A.
Claim 12:
The combination of Aggarwal, Udhayanan, Dammu and Park failed to disclose a text prior loss, however, Zhao Abstract and teachings of “the text perception loss includes text gradient prior loss and text prior loss, the text gradient prior loss adopts L1 norm to constrain the difference between the gradient field of the restored text image and the gradient field of the original image, and the text gradient prior loss” discloses this feature. It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention to have further modified Aggarwal, Udhayanan and Park to include image processing having text prior loss, modifying images quickly and more efficiently, as taught by Zhao. Additionally, the references, Aggarwal, Udhayanan, Dammu Park and Zhao, teaches features that are directed to analogous art and they are directed to the same field on endeavor, such as, image processing. This close relation between the references highly suggests an expectation of success.
As modified:
The combination of Aggarwal, Udhayanan, Dammu, Park and Zhao discloses wherein training, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) using the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) and the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) comprises applying, by the computing system, a text prior loss (See Zhao Abstract and teachings of “the text perception loss includes text gradient prior loss and text prior loss, the text gradient prior loss adopts L1 norm to constrain the difference between the gradient field of the restored text image and the gradient field of the original image, and the text gradient prior loss”) between the text embedding (See Park Paragraph 0047) generated by the text encoder (See Aggarwal Paragraph 0031) for the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) and a reference text embedding (See Park Paragraph 0047) generated for the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) by a reference version of the text encoder (See Aggarwal Paragraph 0031).
Claim 13:
The combination of Aggarwal, Udhayanan, Dammu Park and Zhao discloses wherein training, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), the image-text embedding model (See Udhayanan Paragraphs 0033-0036; 0064-0065) using the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) and the text prompt (See Udhayanan Paragraphs 0034; 0064-0065) comprises applying, by the computing system (See Aggarwal Figure 1, Item 105; Paragraph 0023), an image prior loss (See Zhao Abstract and teachings of “the text perception loss includes text gradient prior loss and text prior loss, the text gradient prior loss adopts L1 norm to constrain the difference between the gradient field of the restored text image and the gradient field of the original image, and the text gradient prior loss”) between the image embedding (See Park Paragraph 0047) generated by the image encoder (See Aggarwal Paragraph 0031) for the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) and a reference image embedding (See Park Paragraph 0047) generated for the modified image (See Aggarwal Figure 2, Item 220; Paragraphs 0035-0041) by a reference version of the image encoder (See Aggarwal Paragraph 0031).
Pertinent Art
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20190108411 relates to the field of image data processing technologies, and, more particularly, to image processing methods and processing devices.
Contact Information
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SHEREE N BROWN whose telephone number is (571)272-4229. The examiner can normally be reached M-F 5:30-2:00 PM EST.
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, SAID BROOME can be reached at (571) 272-2931. 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.
/SHEREE N BROWN/Primary Examiner, Art Unit 2612 March 26, 2026
1 Aggarwal’s Abstract discloses a system that adjusts an image based on a user-input source color and target color.
2 See Udhayanan Paragraph 0034 teachings of “the reference image 202 is a photograph of a dog, the modification text 204 reads “have the dog wear a sweater,” and the target image 206 is a photograph of a dog wearing a sweater”.
3 Udhayanan Paragraphs 0034; 0064-0065 recites “obtaining a reference image 202, a modification text 204, and a target image 206. For example, the reference image 202 is a photograph of a dog, the modification text 204 reads “have the dog wear a sweater,” and the target image 206 is a photograph of a dog wearing a sweater.”
Prosecution Insights