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 .
Response to Amendment
Applicant’s amendments filed on 15 April 2026 have been entered. Claims 1, 8 and 15 have been amended. Claims 1-20 are still pending in this application, with claims 1, 8 and 15 being independent.
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) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Park et al. (US 20210158561 A1), referred herein as Park in view of LE CLERC et al. (US 20170178306 A1), referred herein as CLERC and Tuomi et al. (US 20210225060 A1), referred herein as Tuomi.
Regarding Claim 1, Park in view of CLERC teaches a method of data compression, the method comprising (Park Abstract: Apparatuses, systems, and techniques estimate a pose of an object based on images generated from a combined image volume; [0353] ROP 2126 includes compression logic to compress depth or color data that is written to memory):
accessing at least one input image before detection of any user viewing interest (Park [0060] set of images is obtained based on a 3-D image volume…The plurality of images may comprise two-dimensional (2-D) images of an object; FIG. 3:300: Modeling system).
Park does not but CLERC teaches
generated offline by static 2D computer generated imagery (CGI) (CLERC [0031] generating, e.g. synthesizing, a first face in a first image, by determining a first occluded part of the first face that is occluded by an occluding object, for example a Head-Mounted Display (HMD); [0034] The first image 11 is for example a still image acquired with a digital still camera; [0515] virtual instruments may include software-defined applications for performing one or more processing operations with respect to imaging data generated by imaging devices)
Park does not but Tuomi teaches
subdividing each input image among the at least one input image into image tiles before detection of any user viewing interest (Tuomi [0045] As shown in block 404, the method 400 includes dividing the image into a plurality of tiles; Park FIG.4:400 Rendering system); and
causing a trained Al model to perform a data fetching prediction process in which the trained Al model extracts an image tile among the image tiles from a memory before detection of any user viewing interest and in which the trained Al model creates at least one extrapolated output image that contains more visual data than a corresponding input image inputted to the trained Al model by inclusion of the image tile extracted from the memory before detection of any user viewing interest (Park [0064] the network design may build a 3-D voxel representation of an object by computing 2-D latent features and projecting them to a canonical 3-D voxel using a deprojection unit. This operation may be interpreted as space carving in latent space. The network may render a novel view by rotating the latent voxel representation to the new view and projecting it into the 2-D image space. Using the projected latent features, a decoder may generate a new view image by predicting the depth map of the object at the query view and assigning color for each pixel by combining corresponding pixel values at different reference views; [0065] to reconstruct and render unseen objects, the network may be trained on random 3-D meshes from a dataset; [0066] a neural network is provided that reconstructs a latent representation of a target object given a limited set of reference views and renders the object from arbitrary viewpoints without additional training; [0516] use machine learning models or other AI to perform one or more processing steps; FIG. 5:500 Pose estimation system; FIG. 18B: 1855A-N; 815: neural network training operations).
CLERC discloses techniques estimating a pose of an object based on images generated from a combined image volume, which 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 Park to incorporate the teachings of CLERC, and apply the imaging data generated by imaging devices and HMD synthesizing method to techniques for estimating a pose of an object based on images generated from a combined image volume.
Doing so would provide a perfect framework for immersive experiences in gaming, virtual reality, movie watching or video conferences.
Tuomi discloses a method of tiled rendering of an image for display, which 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 Park to incorporate the teachings of Tuomi, and apply the tile-based image rendering to techniques for estimating a pose of an object based on images generated from a combined image volume.
Doing so would reduce the network overhead.
Regarding Claim 2, Park in view of CLERC and Tuomi teaches the method of claim 1, and further teaches further comprising: presenting the at least one extrapolated output image that contains more visual data than the corresponding input image by inclusion of the image tile extracted from the memory before detection of any user viewing interest (Park [0068] In accordance with the foregoing, in at least one embodiment, the described techniques provide an end-to-end system or process for novel view reconstructions and pose estimation for a target object, where the target object may be selected for engagement by a robotic system. In at least one embodiment, a reconstruction pipeline of the system or process may obtain a collection of reference images as input and generate a flexible representation which can be rendered from novel viewpoints; [0318] a tiling unit 1858 to accelerate tiling operations for tile-based rendering, in which rendering operations for a scene are subdivided in image space, for example to exploit local spatial coherence within a scene or to optimize use of internal caches).
Regarding Claim 3, Park in view of CLERC and Tuomi teaches the method of claim 1, and further teaches wherein: the at least one input image is generated offline by the static 2D CGI prior to the AI model performing the data fetching prediction process that extracts the image tile from the memory before detection of any user viewing interest (Park [0065] to reconstruct and render unseen objects, the network may be trained on random 3-D meshes from a dataset, such as a ShapeNet dataset, that may be textured using images from a dataset, such as a MS-COCO dataset, under different lighting conditions).
Regarding Claim 4, Park in view of CLERC and Tuomi teaches the method of claim 1, and further teaches wherein: the at least one input image is generated offline by the static 2D CGI prior to the AI model creating the at least one extrapolated output image that contains more visual data than the corresponding input image by inclusion of the image tile extracted from the memory before detection of any user viewing interest (Park [0098] In at least one embodiment, at block 606, the feature volumes generated at block 604 are fused or combined to generate a combined feature volume. In at least one embodiment, the combined feature volume is a canonical feature volume. In at least one embodiment, the combined feature volume is a 3-D feature volume. In at least one embodiment, the combined feature volume is generated by the modeling system 300. For example, in at least one embodiment, the block 606 generate the combined feature volume 336).
Regarding Claim 5, Park in view of CLERC and Tuomi teaches the method of claim 1, and further teaches wherein: the data fetching prediction process extracts the image tile from a local cache memory provided by a content delivery network before detection of any user viewing interest (Tuomi [0021] a network connection (e.g., a wireless local area network card for transmission and/or reception of wireless IEEE 802 signals); 7. The method of claim 6, wherein executing the coarse level tiling and the fine level tiling, via the same fixed function hardware, further comprises: executing the coarse level tiling to determine in which tile each of a plurality of primitives is located; and execute the fine level tiling by utilizing local cache memory to accumulate a batch of primitives and render primitives one fine tile at a time.).
Regarding Claim 6, Park in view of CLERC and Tuomi teaches the method of claim 1, and further teaches wherein: the AI model creates the at least one extrapolated output image that contains more visual data than a corresponding input image by generating at least one future tile based on the at least one input image before detection of any user viewing interest (Park [0073] using the plurality of images 302-306 with associated object poses and object segmentation binary masks 308-312, the system 300 may generate a representation of the object which can be rendered with arbitrary camera parameters. The object may be represented as a latent 3-D voxel grid. In at least one embodiment, the representation can be directly manipulated using standard 3-D transformations and enable novel view rendering).
Regarding Claim 7, Park in view of CLERC and Tuomi teaches the method of claim 1, and further teaches wherein: the AI model creates the at least one extrapolated output image that contains more visual data than a corresponding input image by including at least one generated future tile in the at least one extrapolated output image before detection of any user viewing interest (Park [0068] process for novel view reconstructions and pose estimation for a target object, where the target object may be selected for engagement by a robotic system. In at least one embodiment, a reconstruction pipeline of the system or process may obtain a collection of reference images as input and generate a flexible representation which can be rendered from novel viewpoints. In at least one embodiment, multi-view consistency may be utilized to construct a latent representation, and the system or process may not require the use of category specific shape priors).
Regarding Claims 8-14, Park in view of CLERC and Tuomi teaches a system (Park Abstract: Apparatuses, systems, and techniques estimate a pose of an object based on images generated from a combined image volume; [0353] ROP 2126 includes compression logic to compress depth or color data that is written to memory).
The metes and bounds of the of the claims substantially correspond to the limitations set forth in claims 1-7; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Regarding Claims 15-20, Park in view of CLERC and Tuomi teaches a non-transitory storage medium comprising instructions that, when executed by one or more microprocessors of a computer, cause the computer to perform operations (Park Abstract: Apparatuses, systems, and techniques estimate a pose of an object based on images generated from a combined image volume; [0353] ROP 2126 includes compression logic to compress depth or color data that is written to memory; [0620] Clause 21. A machine-readable medium having stored thereon a set of instructions, which if performed by one or more processors, cause the one or more processors to).
The metes and bounds of the of the claims substantially correspond to the limitations set forth in claims 1-6; thus they are rejected on similar grounds and rationale as their corresponding limitations.
Response to Arguments
Applicant’s arguments, see page 7, filed on 15 April 2026, with respect to double patenting rejection have been fully considered and are persuasive. The double patenting rejection of 22 January 2026 has been withdrawn.
Applicant's arguments filed on 15 April 2026, with respect to the 103 rejection have been fully considered but they are not persuasive.
On page 9, Applicant's Remarks, with respect to claims 1, 8 and 15, the applicant argues Park does not teach image tile nor before detection of user viewing interest. The Examiner respectfully disagrees with the first argument. Paragraph [0318] of Park explicitly disclosed tile-based rendering system, and the rendering process is happed at FIG. 4: 400, which is happened before he pose estimation system 500. Therefore, the process of combination with Tuomi is before the detection of user viewing direction. Regarding the first argument, it is respectfully noted that the prior art teaches accessing at least one input image generated offline by static 2D computer generated imagery (CGI) before detection of any user viewing interest, as claimed.
On page 9, Applicant's Remarks, with respect to claims 1, 8 and 15, the applicant argues inclusion of tile before detection is not “necessarily present” in assigning colors and depth map. The Examiner respectfully disagrees with the second argument. As mentions above, the image tiling is part of the rendering processing. However, the claimed limitation of “subdividing each input image among the at least one input image into image tiles” does not exclude other rendering process, such as depth map predicting and color assignment and the rendering process as a whole is proceeded before he pose estimation process. Regarding the second argument, it is respectfully noted that the prior art teaches subdividing each input image among the at least one input image into image tiles before detection of any user viewing interest, as claimed.
On page 10, Applicant's Remarks, with respect to claims 1, 8 and 15, the applicant argues inclusion of tile before detection is not “necessarily present” in traning a network or in rendering unseen object. The Examiner respectfully disagrees with the third argument. As mentions above, Paragraph [0318] of Park explicitly disclosed tile-based rendering system. Park further teaches a neural network is provided that reconstructs a latent representation of a target object given a limited set of reference views and renders the object from arbitrary viewpoints without additional training (see Par. [0066], [0318]-[0319]; FIG. 18A & B). Regarding the third argument, it is respectfully noted that the prior art teaches causing a trained Al model to perform a data fetching prediction process in which the trained Al model extracts an image tile among the image tiles from a memory before detection of any user viewing interest, as claimed.
Conclusion
THIS ACTION IS MADE FINAL. Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to 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 at (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