CTNF 18/677,162 CTNF 101971 Notice of Pre-AIA or AIA Status 07-03-aia AIA 15-10-aia The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA. Priority 02-26 AIA Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55. 12-151 AIA 26-51 12-51 Status of Claims This communication is in response to the Application Filed on 05/29/2024. Claims 1-18 are pending in this application. 07-30-03-h AIA Claim Interpretation 07-30-03 AIA The following is a quotation of 35 U.S.C. 112(f): (f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof. The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph: An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof. 07-30-05 The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked. As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph: (A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function; (B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and (C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function. Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function. Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function. Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are: “a generation device” in claim 10 “a learning device” in claim 10 “a scaling device” in claim 12 Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof. Claim 10: ‘a generation device’ corresponds to Fig. 8 – element 110. “Each of the generation device 110, the learning device 120, and the scaling device 130 may correspond to a processor”, Applicant Specification ¶ [00113]. Claim 10: ‘a learning device’ corresponds to Fig. 8 – element 120. “Each of the generation device 110, the learning device 120, and the scaling device 130 may correspond to a processor”, Applicant Specification ¶ [00113]. Claim 12: ‘a scaling device’ corresponds to Fig. 8 – element 130. “Each of the generation device 110, the learning device 120, and the scaling device 130 may correspond to a processor”, Applicant Specification ¶ [00113]. Claims 11 and 13-18 inherently invoke 112f due to their dependency from claim 10. Claims 1-9 invoke 112f due to “a generation device,” “a learning device,” and “a scaled device” recited therein. If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. Claim Rejections - 35 USC § 103 07-06 AIA 15-10-15 In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status. 07-20-aia AIA 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. 07-23-aia AIA The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows: 1. Determining the scope and contents of the prior art. 2. Ascertaining the differences between the prior art and the claims at issue. 3. Resolving the level of ordinary skill in the pertinent art. 4. Considering objective evidence present in the application indicating obviousness or nonobviousness. 07-21-aia AIA Claim s 1-2, 6-7, 9-11, 15-16, and 18 are rejected under 35 U.S.C. 103 as being unpatentable over Lu et al. (Multi-Task Learning for Single Image Depth Estimation and Segmentation Based on Unsupervised Network, 2020, hereafter, “Lu”) in view of Mousavian et al. (Joint Semantic Segmentation and Depth Estimation with Deep Convolutional Networks, 2016, hereafter, “Mousavian”) . Regarding claim 1 , Lu discloses a multi-task learning method (See Lu, Pg. 10788, right col., par. 1, lines 1-2, multi-task learning framework), the method comprising: PNG media_image1.png 468 1062 media_image1.png Greyscale generating, by a generation device, a first feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) based on a first input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input) through a multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ); generating, by the generation device, a first output image (See Lu, Fig. 4; Pg. 10793, right col., section C, line 9, image segmentation output) based on the first feature (See Lu, Pg. 10789, left col., section image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) through a first decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. Examiner considers a respective decoder of each task as there being a first decoder for a first task ) for a first task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers segmentation as “first task” ); [generating, by the generation device, a first loss] based on the first output image (See Lu, Fig. 4; Pg. 10793, right col., section C, line 9, image segmentation output) [and a first ground truth (GT)] for the first task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers segmentation as “first task” ); generating, by the generation device, a second feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images) based on the first input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input) through a pretrained first encoder (See Lu, Pg. 10789, right col., section III, lines 18-19, separate encoder for each task to explore their unique feature. Examiner considers this to mean an encoder exploring, or “pretrained”, for a specific task ) for a second task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers depth estimation as “second task” ); generating, by the generation device, a second loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel) based on the first feature ( See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) and the second feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images) ; and learning, by a learning device, the multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ) and the first decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. Examiner considers a respective decoder of each task as there being a first decoder for a first task ) [based on the first loss] and the second loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel). However, Lu fails to disclose generating, by the generation device, a first loss [based on the first output image] and a first ground truth (GT) [for the first task] and [learning, by a learning device, the multi-task and the first decoder] based on the first loss [and the second loss.] Mousavian teaches generating, by the generation device, a first loss (See Mousavian, Pg. 3, right col., section 3.2, Equation 2 for Semantic loss) [based on the first output image] and a first ground truth (GT) (See Mousavian, Pg. 3, right col., section 3.2, , Equation 2 for Semantic loss, line 7, where C i * is the ground truth label of pixel i ) [for the first task] and [learning, by a learning device, the multi-task and the first decoder] based on the first loss (See Mousavian, Pg. 3, right col., section 3.2, Equation 2 for Semantic loss) [and the second loss.] Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to generate a first loss based on the first output image and a first ground truth (GT) for the first task and learn the multi-task and the first decoder based on the first loss and the second loss based on the method of Mousavian’s reference. The suggestion/motivation would have been for feasibility of training parts of the model for each task and then fine tuning the full, combined model on both tasks to improve the accuracy of the final results as suggested by Mousavian in the Abstract. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Mousavian with Lu to obtain the invention as specified in claim 1. Regarding claim 2 , Lu and Mousavian disclose the method of claim 1, further comprising: generating, by the generation device, a third feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images. Examiner considers third feature and second features to be similar ) based on a second input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input. Examiner considers first input image and second input image to be similar ) through the multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ); generating, by the generation device, a second output image (See Lu, Pg. 10790, left col., section A, lines 24-25, the decoder further constrains the output disparity to produce the wrapped image) based on the third feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images. Examiner considers third feature and second features to be similar ) through a second decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. E xaminer considers a respective decoder of each task as there being a second decoder for a second task ) for the second task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers depth estimation as “second task” ) ; [generating, by the generation device, a third loss] based on the second output image (See Lu, Pg. 10790, left col., section A, lines 24-25, the decoder further constrains the output disparity to produce the wrapped image) [and a second GT] for the second task (See Lu, Pg. 10789, Figure 2, two tasks (single image depth estimation and segmentation. Examiner considers depth estimation as “second task” ) ; generating, by the generation device, a fourth feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering. Examiner considers the first feature and fourth feature to be similar ) based on the second input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input. Examiner considers first input image and second input image to be similar ) through a pretrained second encoder (See Lu, Pg. 10789, right col., section III, lines 18-19, separate encoder for each task to explore their unique feature. Examiner considers this to mean an encoder exploring, or “pretrained”, for a specific task ) for the first task (See Lu, Pg. 10789, Figure 2, two tasks (single image depth estimation and segmentation. Examiner considers segmentation as “first task” ) ; generating, by the generation device, a fourth loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel) based on the third feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images. Examiner considers third feature and second features to be similar ) and the fourth feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering. Examiner considers the first feature and fourth feature to be similar ) ; and learning, by the learning device, the multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ); and the second decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. E xaminer considers a respective decoder of each task as there being a second decoder for a second task ) [based on the third loss] and the fourth loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to generate a first loss based on the first output image and a first ground truth (GT) for the first task and learn the multi-task and the first decoder based on the first loss and the second loss based on the method of Mousavian’s reference. The suggestion/motivation would have been for feasibility of training parts of the model for each task and then fine tuning the full, combined model on both tasks to improve the accuracy of the final results as suggested by Mousavian in the Abstract. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Mousavian with Lu to obtain the invention as specified in claim 2. Regarding claim 6 , Lu and Mousavian teach the method of claim 1, wherein the pretrained first encoder for the second task is an encoder with a structure that is the same as a structure of the multi-task encoder (See Lu, Pg. 10789, right col., section III, lines 18-19, applies a separate encoder for each task to explore their unique features, and then share them; Pg. 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task and the separate encoder is exploring unique features of a task in which it is then shared, similar to the multi-task encoder). Regarding claim 7 , Lu and Mousavian teach the method of claim 1, wherein the first task is an image segmentation task, and wherein the second task is a depth estimation task (See Lu, Pg. 10788, right col., lines 1-3, multi-task learning framework for simultaneous single image depth estimation and image segmentation. Examiner considers image segmentation as the first task and depth estimation as the second task ). Regarding claim 9 , Lu and Mousavian teach the method of claim 1, wherein the learning of the multi-task encoder and the first decoder includes: learning, by the learning device, the multi-task encoder and the first decoder while fixing a parameter of the pretrained first encoder (See Lu, Pg. 10790, left col., lines 7-10, adding the segmentation context to the depth encoder, we can obtain more accurate feature representation than just focusing on depth task. The same for encoding depth context into segmentation task. Examiner considers adding segmentation context into the depth encoder and vice versa as the encoder/decoder network learning from each other). Regarding claim 10 , A multi-task learning apparatus (See Lu, Pg. 10788, right col., par. 1, lines 1-2, multi-task learning framework; Pg. 10792, left col., 11-12, framework is implemented in PyTorch) comprising: a memory configured to store computer-executable instructions (See Lu, Pg. 10788, right col., 8-11, unsupervised CNN framework consisting of loss constraints from both spatial and spectral perspectives to simultaneously train the neural network for each task. Examiner considers an unsupervised neural network with means to train itself as holding some sort of memory for learning purposes ); a generation device including a first processor configured to access the memory and to execute the computer-executable instructions (See Lu, Pg., 10792, left col., lines 12-13, NVIDIA 1080ti GPU) , wherein the generation device is configured to generate a first feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) based on a first input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input) by using a multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ), generate a first output image (See Lu, Fig. 4; Pg. 10793, right col., section C, line 9, image segmentation output) based on the first feature (See Lu, Pg. 10789, left col., section image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) by using a first decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. Examiner considers a respective decoder of each task as there being a first decoder for a first task ) for a first task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers segmentation as “first task” ), [generate a first loss] based on the first output image (See Lu, Fig. 4; Pg. 10793, right col., section C, line 9, image segmentation output) [and a GT] for the first task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers segmentation as “first task” ), generate a second feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images) based on the first input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input) by using a pretrained first encoder (See Lu, Pg. 10789, right col., section III, lines 18-19, separate encoder for each task to explore their unique feature. Examiner considers this to mean an encoder exploring, or “pretrained”, for a specific task ) for a second task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers depth estimation as “second task” ), and generate a second loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel) based on the first feature ( See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) and the second feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images) ; and a learning device including a second processor configured to access the memory and to execute the computer-executable instructions (See Lu, Pg., 10792, left col., lines 12-13, NVIDIA 1080ti GPU. Examiner considers the first processor and second processor to be similar ) , wherein the learning device is configured to learn the multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ) and the first decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. Examiner considers a respective decoder of each task as there being a first decoder for a first task ) [based on the first loss] and the second loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel). However, Lu fails to disclose generating, by the generation device, a first loss [based on the first output image] and a first ground truth (GT) [for the first task] and [learning, by a learning device, the multi-task and the first decoder] based on the first loss [and the second loss.] Mousavian teaches generate a first loss (See Mousavian, Pg. 3, right col., section 3.2, Equation 2 for Semantic loss) [based on the first output image] and a GT (See Mousavian, Pg. 3, right col., section 3.2, , Equation 2 for Semantic loss, line 7, where C i * is the ground truth label of pixel i ) [for the first task] and [wherein the learning device is configured to learn the multi-task encoder and the first decoder] based on the first loss (See Mousavian, Pg. 3, right col., section 3.2, Equation 2 for Semantic loss) [and the second loss.] Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to generate a first loss based on the first output image and a first ground truth (GT) for the first task and learn the multi-task and the first decoder based on the first loss and the second loss based on the method of Mousavian’s reference. The suggestion/motivation would have been for feasibility of training parts of the model for each task and then fine tuning the full, combined model on both tasks to improve the accuracy of the final results as suggested by Mousavian in the Abstract. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Mousavian with Lu to obtain the invention as specified in claim 10. Regarding claim 11 , Lu and Mousavian disclose the multi-task learning apparatus of claim 10, wherein the generation device is configured to generate a third feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images. Examiner considers third feature and second features to be similar ) based on a second input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input. Examiner considers first input image and second input image to be similar ) by using the multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ), generate a second output image (See Lu, Pg. 10790, left col., section A, lines 24-25, the decoder further constrains the output disparity to produce the wrapped image) based on the third feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images. Examiner considers third feature and second features to be similar ) by using a second decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. E xaminer considers a respective decoder of each task as there being a second decoder for a second task ) for the second task (See Lu, Pg. 10789, Figure 2, The proposed architecture consists of two tasks (single image depth estimation and segmentation). Examiner considers depth estimation as “second task” ) , [generate a third loss] based on the second output image (See Lu, Pg. 10790, left col., section A, lines 24-25, the decoder further constrains the output disparity to produce the wrapped image) [and a second GT] for the second task (See Lu, Pg. 10789, Figure 2, two tasks (single image depth estimation and segmentation. Examiner considers depth estimation as “second task” ) , generate a fourth feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering. Examiner considers the first feature and fourth feature to be similar ) based on the second input image (See Lu, Pg. 10793, right col., section VI, line 3, one RGB image as input. Examiner considers first input image and second input image to be similar ) by using a pretrained second encoder (See Lu, Pg. 10789, right col., section III, lines 18-19, separate encoder for each task to explore their unique feature. Examiner considers this to mean an encoder exploring, or “pretrained”, for a specific task ) for the first task (See Lu, Pg. 10789, Figure 2, two tasks (single image depth estimation and segmentation. Examiner considers segmentation as “first task” ) , and generate a fourth loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel) based on the third feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images. Examiner considers third feature and second features to be similar ) and the fourth feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering. Examiner considers the first feature and fourth feature to be similar ) , and the learning device is configured to learn the multi-task encoder (See Lu, page 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task ) and the second decoder (See Lu, Pg. 10789, Figure 2, The shared encoder is connected to the respective decoder of each task to produce a pixel-wise depth map and segmentation. E xaminer considers a respective decoder of each task as there being a second decoder for a second task ) [based on the third loss] and the fourth loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to generate a first loss based on the first output image and a first ground truth (GT) for the first task and learn the multi-task and the first decoder based on the first loss and the second loss based on the method of Mousavian’s reference. The suggestion/motivation would have been for feasibility of training parts of the model for each task and then fine tuning the full, combined model on both tasks to improve the accuracy of the final results as suggested by Mousavian in the Abstract. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Mousavian with Lu to obtain the invention as specified in claim 11. Regarding claim 15 , Lu and Mousavian teach the multi-task learning apparatus of claim 10, wherein the first encoder pretrained for the second task is an encoder with a structure that is the same as a structure of the multi-task encoder (See Lu, Pg. 10789, right col., section III, lines 18-19, applies a separate encoder for each task to explore their unique features, and then share them; Pg. 10789, right col., section III, correlate these two tasks by sharing weights in their encoder network. Examiner considers this as a “multi-task encoder” as it is receiving information from more than one task and the separate encoder is exploring unique features of a task in which it is then shared, similar to the multi-task encoder). Regarding claim 16 , Lu and Mousavian teach the multi-task learning apparatus of claim 10, wherein the first task is an image segmentation task, and wherein the second task is a depth estimation task (See Lu, Pg. 10788, right col., lines 1-3, multi-task learning framework for simultaneous single image depth estimation and image segmentation. Examiner considers image segmentation as the first task and depth estimation as the second task ). Regarding claim 18 , Lu and Mousavian teach the multi-task learning apparatus of claim 10, wherein the learning device is configured to learn the multi-task encoder and the first decoder while fixing a parameter of the pretrained first encoder (See Lu, Pg. 10790, left col., lines 7-10, adding the segmentation context to the depth encoder, we can obtain more accurate feature representation than just focusing on depth task. The same for encoding depth context into segmentation task. Examiner considers adding segmentation context into the depth encoder and vice versa as the encoder/decoder network learning from each other) . 07-21-aia AIA Claim s 3-4 and 12-13 are rejected under 35 U.S.C. 103 as being unpatentable over Lu et al. (Multi-Task Learning for Single Image Depth Estimation and Segmentation Based on Unsupervised Network, 2020, hereafter, “Lu”) in view of Mousavian et al. (Joint Semantic Segmentation and Depth Estimation with Deep Convolutional Networks, 2016, hereafter, “Mousavian”), and further in view of Wang et al. (Semi-supervised Multi-task Learning for Semantics and Depth, 2022, hereafter, “Wang”) . Regarding claim 3 , Lu and Mousavian teach the method of claim 1, wherein the generating of the second loss includes: [scaling, by a scaling device, the first feature to correspond to the second feature;] and generating, by the generation device, the second loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel) [based on the scaled first feature] and the second feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images) . However, Lu and Mousavian fail to teach scaling, by a scaling device, the first feature to correspond to the second feature; and [generating, by the generation device, the second loss] based on the scaled first feature and [the second feature.] Wang teaches scaling, by a scaling device, the first feature to correspond to the second feature (See Wang, Pg 2667, right col, section 3.4, lines 12-16, The depth decoder is constructed with several convolutional layers and up-sampling operations to produce detailed depth features and a regression layer to estimate depth. Finally, we apply an up-sampling layer to the output maps for both tasks to match the input image size. Examiner considers the up-sampling layer used on both tasks to get a matching size the same as “scaling” or matching features ); and generating, by the generation device, the second loss] based on the scaled first feature (See Wang, Pg 2667, right col, section 3.4, lines 12-16, The depth decoder is constructed with several convolutional layers and up-sampling operations to produce detailed depth features and a regression layer to estimate depth. Finally, we apply an up-sampling layer to the output maps for both tasks to match the input image size. Examiner considers the up-sampling layer used on both tasks to get a matching size the same as “scaling” or matching features ) [and the second feature.] Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to scale the first feature to correspond to the second feature based on the method of Wang’s reference. The suggestion/motivation would have been for the generalization ability by sharing representations among related tasks as well as the benefit of multiple sources with supervision as suggested by Wang in the Introduction. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Wang with Lu and Mousavian to obtain the invention as specified in claim 3. Regarding claim 4 , Lu, Mousavian, and Wang teach the method of claim 3, wherein the scaling of the first feature includes: scaling, by the scaling device, the first feature by inputting the first feature into a convolution layer (See Wang, Pg. 2667, right col., section 3.4, lines 12-16, The depth decoder is constructed with several convolutional layers and up-sampling operations to produce detailed depth features and a regression layer to estimate depth. Finally, we apply an up-sampling layer to the output maps for both tasks to match the input image size. Decoders are made up of convolution layers which are used for feature matching ). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to scale the first feature by inputting the first feature into a convolution layer based on the method of Wang’s reference. The suggestion/motivation would have been for the generalization ability by sharing representations among related tasks as well as the benefit of multiple sources with supervision as suggested by Wang in the Introduction. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Wang with Lu and Mousavian to obtain the invention as specified in claim 4. Regarding claim 12 , Lu and Mousavian teach the multi-task learning apparatus of claim 10, further comprising [a scaling device] including a third processor configured to access the memory and to execute the computer-executable instructions (See Lu, Pg., 10792, left col., lines 12-13, NVIDIA 1080ti GPU. Examiner considers first processor and third processor to be similar ) , [wherein the scaling device is configured to scale the first feature to correspond to the second feature;] and wherein the generation device is configured to generate the second loss (See Lu, Pg 10790, right col., section IV, Equation 3. This equation takes into consideration segmentation loss and disparity loss; Pg 10791, left col., section B, lines 1-5, The disparity smoothness loss embeds the original image gradient change information to enforce the disparity maps to be smooth. Examiner considers gradient to be an intensity feature which is cited above for “first feature”; Pg 10791, right col., section D, lines 1-3, the segmentation loss is to obtain a more accurate cluster label for each input pixel) [based on the scaled first feature] and the second feature (See Lu, Pg. 10789, right col., section III, lines 10-11, extracts essential geometry features from input images) . However, Lu and Mousavian fail to teach a scaling device [including a third processor configured to access the memory and to execute the computer-executable instructions,] wherein the scaling device is configured to scale the first feature to correspond to the [second feature.] Wang teaches a scaling device [including a third processor configured to access the memory and to execute the computer-executable instructions,] wherein the scaling device is configured to scale the first feature (See Wang, Pg 2667, right col, section 3.4, lines 12-16, The depth decoder is constructed with several convolutional layers and up-sampling operations to produce detailed depth features and a regression layer to estimate depth. Finally, we apply an up-sampling layer to the output maps for both tasks to match the input image size. Examiner considers the up-sampling layer used on both tasks to get a matching size the same as “scaling” or matching features ) to correspond to the [second feature;] and generating, by the generation device, the second loss] based on the scaled first feature (See Wang, Pg 2667, right col, section 3.4, lines 12-16, The depth decoder is constructed with several convolutional layers and up-sampling operations to produce detailed depth features and a regression layer to estimate depth. Finally, we apply an up-sampling layer to the output maps for both tasks to match the input image size. Examiner considers the up-sampling layer used on both tasks to get a matching size the same as “scaling” or matching features ) [and the second feature.] Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to scale the first feature to correspond to the second feature based on the method of Wang’s reference. The suggestion/motivation would have been for the generalization ability by sharing representations among related tasks as well as the benefit of multiple sources with supervision as suggested by Wang in the Introduction. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Wang with Lu and Mousavian to obtain the invention as specified in claim 12. Regarding claim 13 , Lu, Mousavian, and Wang teach the multi-task learning apparatus of claim 12, wherein the scaling device is configured to scale the first feature by inputting the first feature into a convolution layer (See Wang, Pg. 2667, right col., section 3.4, lines 12-16, The depth decoder is constructed with several convolutional layers and up-sampling operations to produce detailed depth features and a regression layer to estimate depth. Finally, we apply an up-sampling layer to the output maps for both tasks to match the input image size. Decoders are made up of convolution layers which are used for feature matching ). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to scale the first feature by inputting the first feature into a convolution layer based on the method of Wang’s reference. The suggestion/motivation would have been for the generalization ability by sharing representations among related tasks as well as the benefit of multiple sources with supervision as suggested by Wang in the Introduction. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Wang with Lu and Mousavian to obtain the invention as specified in claim 13 . 07-21-aia AIA Claim s 5 and 14 are rejected under 35 U.S.C. 103 as being unpatentable over Lu et al. (Multi-Task Learning for Single Image Depth Estimation and Segmentation Based on Unsupervised Network, 2020, hereafter, “Lu”) in view of Mousavian et al. (Joint Semantic Segmentation and Depth Estimation with Deep Convolutional Networks, 2016, hereafter, “Mousavian”), and further in view of Wang et al. (Semi-supervised Multi-task Learning for Semantics and Depth, 2022, hereafter, “Wang”), and Hu et al. (Deep Depth Completion From Extremely Sparse Data: A Survey, 2023, hereafter, “Hu”) . Regarding claim 5 , Lu, Mousavian, and Wang teach the method of claim 3, wherein the generating of the second loss includes: generating, by the generation device, the second loss by [calculating a distance between the scaled first feature and the second feature.] However, Lu, Mousavian, and Wang fail to teach [generating, by the generation device, the second loss by] calculating a distance between the scaled first feature and the second feature. Hu teaches [generating, by the generation device, the second loss by] calculating a distance between the scaled first feature and the second feature (See Hu, Pg. 8248, right col., section 4.1.2, par. 2, prior distance map between depth points based on a Euclidean distance. Euclidean distance is being used to find the distance between points, in which the points can be any type, such as for two different features). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu‘s reference to generate the second loss by calculating a distance between the scaled first feature and the second feature based on the method of Hu’s reference. The suggestion/motivation would have been to provide accurate and robust distance measurements with true scene scales as suggested by Hu in the Abstract . Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Hu with Lu, Mousavian, and Wang to obtain the invention as specified in claim 5. Regarding claim 14 , Lu, Mousavian, and Wang teach the multi-task learning apparatus of claim 12, wherein the generation device is configured to generate the second loss by [calculating a distance between the scaled first feature and the second feature.] However, Lu, Mousavian, and Wang fail to teach [the multi-task learning apparatus of claim 12, wherein the generation device is configured to generate the second loss by] calculating a distance between the scaled first feature and the second feature. Hu teaches wherein the generation device is configured to generate the second loss by calculating a distance between the scaled first feature and the second feature (See Hu, Pg. 8248, right col., section 4.1.2, par. 2, prior distance map between depth points based on a Euclidean distance. Euclidean distance is being used to find the distance between points, in which the points can be any type, such as for two different features). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu‘s reference to generate the second loss by calculating a distance between the scaled first feature and the second feature based on the method of Hu’s reference. The suggestion/motivation would have been to provide accurate and robust distance measurements with true scene scales as suggested by Hu in the Abstract . Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Hu with Lu, Mousavian, and Wang to obtain the invention as specified in claim 14 . 07-21-aia AIA Claim s 8 and 17 are rejected under 35 U.S.C. 103 as being unpatentable over Lu et al. (Multi-Task Learning for Single Image Depth Estimation and Segmentation Based on Unsupervised Network, 2020, hereafter, “Lu”) in view of Mousavian et al. (Joint Semantic Segmentation and Depth Estimation with Deep Convolutional Networks, 2016, hereafter, “Mousavian”) and further in view of Kumar et al. (OmniDet: Surround View Cameras based Multi-task Visual Perception Network for Autonomous Driving, 2023, hereafter, “Kumar”) . Regarding claim 8 , Lu and Mousavian teach the method of claim 1, wherein the generating of the first output image includes: generating, by the generation device, the first output image (See Lu, Fig. 4; Pg. 10793, right col., section C, line 9, image segmentation output) based on the first feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) and [camera calibration information.] However, Lu and Mousavian fail to teach [generating, by the generation device, the first output image based on the first feature] and camera calibration information. Kumar teaches [generating, by the generation device, the first output image based on the first feature] and camera calibration information (See Kumar, Pg 3, right col., section A, normalized coordinates per pixel are used for these channels by incorporating information from the camera calibration. We concatenate these tensors and represent them by Ct and pass it along with the input features to our SAN pairwise and patchwise operation modules). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to generate the first output image based on the first feature and camera calibration information based on the method of Kumar’s reference. The suggestion/motivation would have been to generate different maps using the camera intrinsic parameters as suggested by Kumar on Pg. 4, right col., section A, par. 3, lines 16-17. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Kumar with Lu and Mousavian to obtain the invention as specified in claim 8. Regarding claim 17 , Lu and Mousavian teach the multi-task learning apparatus of claim 10, wherein the generation device is configured to generate the first output image (See Lu, Fig. 4; Pg. 10793, right col., section C, line 9, image segmentation output) based on the first feature (See Lu, Pg. 10789, left col., section: image segmentation, lines 2-3, features such as color, intensity or textures to realize pixel-wise clustering) and [camera calibration information.] However, Lu and Mousavian fail to teach [wherein the generation device is configured to generate the first output image based on the first feature] and camera calibration information. Kumar teaches [wherein the generation device is configured to generate the first output image based on the first feature] and camera calibration information (See Kumar, Pg 3, right col., section A, normalized coordinates per pixel are used for these channels by incorporating information from the camera calibration. We concatenate these tensors and represent them by Ct and pass it along with the input features to our SAN pairwise and patchwise operation modules). Thus, it would have been obvious to one of ordinary skills in the art before the effective filing date of the claimed invention to modify Lu’s reference to generate the first output image based on the first feature and camera calibration information based on the method of Kumar’s reference. The suggestion/motivation would have been to generate different maps using the camera intrinsic parameters as suggested by Kumar on Pg. 4, right col., section A, par. 3, lines 16-17. Further, one skilled in the art could have combined the elements as described above by known method with no change in their respective functions, and the combination would have yielded nothing more than predictable results. Therefore, it would have been obvious to combine Kumar with Lu and Mousavian to obtain the invention as specified in claim 17 . Conclusion 07-96 AIA The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Zhang et al. (US 20230326028 A1) discloses a method for using machine learning models to obtain a redefined depth map of digital images using segmentation masks to aid in the process. Zama et al. (Geometry meets semantics for semi-supervised monocular depth estimation, 2018) discloses of method of using a deep learning approach for joint semantic segmentation and depth estimation in which learning semantic information also improves depth estimation. June et al. (Optimal Configuration of Multi-Task Learning for Autonomous Driving, 2023) discloses an optimal neural network configuration for segmentation, depth estimation, and lane detection and implementation of a multi-task decision and optimization algorithm. Any inquiry concerning this communication or earlier communications from the examiner should be directed to Jasmin Marcelino Hernandez whose telephone number is (571) 270-0211. The examiner can normally be reached 7am-3pm 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, Henok Shiferaw can be reached at (571) 272-4637. 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. /JASMIN MARCELINO HERNANDEZ/Examiner, Art Unit 2676 /AMANDEEP SAINI/Supervisory Patent Examiner, Art Unit 2662 Application/Control Number: 18/677,162 Page 2 Art Unit: 2676 Application/Control Number: 18/677,162 Page 3 Art Unit: 2676 Application/Control Number: 18/677,162 Page 4 Art Unit: 2676 Application/Control Number: 18/677,162 Page 5 Art Unit: 2676 Application/Control Number: 18/677,162 Page 6 Art Unit: 2676 Application/Control Number: 18/677,162 Page 7 Art Unit: 2676 Application/Control Number: 18/677,162 Page 8 Art Unit: 2676 Application/Control Number: 18/677,162 Page 9 Art Unit: 2676 Application/Control Number: 18/677,162 Page 10 Art Unit: 2676 Application/Control Number: 18/677,162 Page 11 Art Unit: 2676 Application/Control Number: 18/677,162 Page 12 Art Unit: 2676 Application/Control Number: 18/677,162 Page 13 Art Unit: 2676 Application/Control Number: 18/677,162 Page 14 Art Unit: 2676 Application/Control Number: 18/677,162 Page 15 Art Unit: 2676 Application/Control Number: 18/677,162 Page 16 Art Unit: 2676 Application/Control Number: 18/677,162 Page 17 Art Unit: 2676 Application/Control Number: 18/677,162 Page 18 Art Unit: 2676 Application/Control Number: 18/677,162 Page 19 Art Unit: 2676 Application/Control Number: 18/677,162 Page 20 Art Unit: 2676 Application/Control Number: 18/677,162 Page 21 Art Unit: 2676 Application/Control Number: 18/677,162 Page 22 Art Unit: 2676 Application/Control Number: 18/677,162 Page 23 Art Unit: 2676 Application/Control Number: 18/677,162 Page 24 Art Unit: 2676 Application/Control Number: 18/677,162 Page 25 Art Unit: 2676 Application/Control Number: 18/677,162 Page 26 Art Unit: 2676 Application/Control Number: 18/677,162 Page 27 Art Unit: 2676 Application/Control Number: 18/677,162 Page 28 Art Unit: 2676 Application/Control Number: 18/677,162 Page 29 Art Unit: 2676 Application/Control Number: 18/677,162 Page 30 Art Unit: 2676
Prosecution Insights