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 Arguments filed 05/26/2026
The applicant’s arguments on page 7-8 regarding the previous 102 rejection are persuasive. Therefore the claims are now rejected under 103 to address the applicant’s arguments and amendment.
Allowable Subject Matter
Claims 4-6 and 13-14 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.
Information Disclosure Statement
The information disclosure statements filed on 05/26/2026 and 06/23/2026 were considered.
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.
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.
Claim(s) 1-3, 7-12, and 15-20 are rejected under 35 U.S.C. 103 as being unpatentable over Bailey et al. (US 20220262073 A1) in view of Le et al. (US 20220292765 A1).
Regarding Claim 1, Bailey teaches a computer-implemented method for fitting a shape model for an object to a set of constraints associated with a target shape , the method comprising: determining, based on the set of constraints (Bailey [0163] “input image”), one or more ground truth positions (Bailey [0163] “facial landmarks”) of one or more points on the target shape (Bailey [0106] “deformed mesh output”); (Bailey [0163] “The facial landmarks are detected on the input image.”) generating, via execution of a set of neural networks, […] fitting parameters (Bailey [0064] “For example, the rig parameters 302 may include inputs for controlling movement in the lip, eye brow, nose, etc.”) associated with […] the one or more points; (Bailey [0163] “To animate the mesh, the movement of the detected landmarks in the recording was tracked and the IK model was used to estimate the rig parameters required to match the new landmark configuration.”) computing, via the shape model, one or more predicted positions (Bailey [0065] “final vertex positions”) of the one or more points based on the set of fitting parameters; (Bailey [0060] “According to at least one embodiment, a method of approximating utilizes artist-created texture coordinates U∈R.sup.n×2 of the facial mesh. The approximation may rely on CNNs, which generate deformation maps given input rig parameters.” [0065] “For each of the deformation maps, vertex offsets may be extracted by performing interpolation of the deformation map at each vertex position in texture coordinate space.” [0076] “With continued reference to FIG. 3, the coarse approximation computes the final vertex positions for the mesh by adding (e.g., at adder 310) the vertex offsets to the vertices of the neutral pose of the mesh”) training the set of neural networks based on one or more losses associated with the one or more predicted positions and the one or more ground truth positions (Bailey [0163] “The rig parameters are then passed to a disclosed approximation model to produce the deformed target mesh.” [0077] “Given the approximation model, a loss function is defined to find the optimal model parameters θ.sub.k. According to at least one embodiment, a loss function is proposed that penalizes both inaccuracies in approximated vertex positions, as well as inaccuracies in face normals on the mesh. Given a target mesh V, and approximated vertex offsets Δ, the loss function may be defined in Equation (1)” [0081] “Because the coarse approximation model works on separate mesh segments, the model could produce discontinuities across boundaries and/or seams between particular mesh segments. To address (e.g., minimize) this potential problem, the error function (e.g., the loss function of Equation (1)) may strongly penalize inaccurate face normals, to encourage smooth (or smoother) results along mesh segment boundaries. Penalizing normal errors also suppresses low-amplitude, high-frequency errors that may be visually disturbing or distracting.” Examiner Note: To penalize losses the predicted has to be compared to ground truth positions.); and generating, via execution of the trained set of neural networks, a three-dimensional (3D) model (Bailey [0185] “mesh deformation”) corresponding to the target shape. (Bailey [0180] “a second plurality of deformation maps is generated by applying a second plurality of neural network-trained models” [0182] “a second plurality of vertex offsets is extracted based on the second plurality of deformation maps.” [0185] “The second plurality of vertex offsets is applied to the neutral mesh by applying the second plurality of vertex offsets to values of at most a subset of the vertices of the neutral mesh to generate the mesh deformation. The subset of the vertices of the neutral mesh may correspond to one or more regions of the neutral mesh exhibiting a level of approximation error that is above a particular threshold.”) but doesn’t explicitly disclose: […] per-point […] each of […].
Le teaches […] per-point […] each of […] (¶13, “Recently, learning-based techniques have been proposed to fit 3D primitives to a 3D point cloud. A recent work proposed a supervised learning-based framework called Supervised Primitive Fitting Network (SPFN) that learns a configuration of 3D primitives that represents a 3D point cloud. Instead of directly regressing representative primitive parameters such as dimensions, angles, or positions of 3D primitives, SPFN first predicts per-point features (association between each point and each primitive, primitive type, and surface normal), and a subsequent differentiable module computes representative primitive parameters by minimizing the fitting error. However, SPFN has limited accuracy and cannot efficiently handle high-resolution 3D point clouds. For example, the ability to process high-resolution 3D point clouds in a neural network with consumer GPUs can be limited by memory. For example, even with a relatively simple 3D point cloud processing architecture such as PointNet, memory limitations during training can limit input 3D point clouds to approximately 10 k points, whereas 3D point cloud scans may include 100 k to one million points. Simply downsampling a 3D point cloud results in the loss of fine detail. As a result, SPFN is not equipped to fit small 3D primitives that represent fine detail from high-resolution 3D point clouds.”).
Therefore it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Bailey in view of Le as Supervised Primitive Fitting Network (SPFN) can robustly detect a varying number of geometric primitives at different scales from 3D point clouds without any manual per-input parameter tuning or user control.
Regarding Claim 2, Bailey in view of Le teaches the computer-implemented method of claim 1, wherein determining the one or more ground truth positions comprises: deforming a template mesh (Bailey [0110] “deformed mesh”) to match the target shape and determining the one or more ground truth positions (Bailey [0110] “landmark vertices”) of the one or more points in the template mesh. (Bailey [0110] “the positions of the landmark vertices … in the deformed mesh were gathered.” Examiner’s Note: The rig function (Bailey [0059]) maps the rig parameters to the deformed mesh, and the rig function is based off the “character's facial rig with a polygonal mesh,” meaning the deformed mesh corresponds to the target shape, the input image.)
Regarding Claim 3, Bailey in view of Le teaches the computer-implemented method of claim 1, wherein the one or more ground truth positions comprise a set of two-dimensional (2D) positions of a set of landmarks associated with an image of the target shape. (Bailey [0163] “The facial landmarks are detected on the input image.” [0034] “While this disclosure is presented in the context of 3D animation applications, it is not limited thereto, and other implementations of the systems, media, and methods described herein are contemplated, including deformation of geometric models within a 2D or 3D coordinate system, as well as for various interactive geometric modeling applications involving production and modification of geometric models, including, but not limited to, rigging, animation, architecture, automotive design, consumer product design, virtual reality applications, augmented reality applications, game development, visual effects, 3D printing, and the like. Any reference in this disclosure to a geometric model or components thereof, within a 3D model or 3D space will be understood to include the disclosure as applicable to 2D models and 2D spaces.” Examiner Note: This embodiment shows these landmarks can be considered in either a 3D or 2D space.)
Regarding Claim 7, Bailey in view of Le teaches the computer-implemented method of claim 1, wherein the one or more losses comprise one or more distances between the one or more predicted positions (Bailey [0076] “With continued reference to FIG. 3, the coarse approximation computes the final vertex positions for the mesh by adding (e.g., at adder 310) the vertex offsets to the vertices of the neutral pose of the mesh 312.” [0077] “Given the approximation model, a loss function is defined to find the optimal model parameters θ.sub.k. According to at least one embodiment, a loss function is proposed that penalizes both inaccuracies in approximated vertex positions, as well as inaccuracies in face normals on the mesh.”)
and the one or more ground truth positions. (Bailey [0066] “According to at least one embodiment, a loss function is proposed that penalizes both inaccuracies in approximated vertex positions, as well as inaccuracies in face normals on the mesh.” Examiner Note: Inaccuracies in the approximated vertex positions (which are a predicted position) are due to a comparison of the distance between the predicted and ground truth positions.)
Regarding Claim 8, Bailey in view of Le teaches the computer-implemented method of claim 1, wherein the one or more losses comprise a coefficient regularization loss associated with a set of blending coefficients included in the set of fitting parameters. (Bailey [0114] “r(p) … may denote the rig function that maps the rig parameters p to the subset of vertices V_C.” [0117] “The loss function used to train the model contains both a point-matching component to ensure that the deformed mesh closely matches the control points, as well as a regularization component to avoid large rig parameters that would create unnatural poses.” Examiner’s Note: The rig parameters (fitting parameters) determine how the mesh is deformed; they act as blending coefficients that control the contribution of deformation influences on the final shape. It is also explicitly stated that there is a loss function that includes a regularization component. Accordingly, the rig parameters function as blending coefficients included in the fitting parameters, and the loss includes a coefficient regularization loss associated with those blending coefficients.)
Regarding Claim 9, Bailey in view of Le teaches the computer-implemented method of claim 1, wherein computing the one or more predicted positions of the one or more points comprises: generating, via the shape model, a set of attributes (Bailey [0060] “deformation maps”) associated with a set of learned shapes for the object (Bailey [0052] “a deformation model is learned without requiring (or otherwise utilizing) assumption of a skeletal system, which is appropriate for complex facial rigs.“) and computing the one or more predicted positions based on the set of attributes and the set of fitting parameters. (Bailey [0052] “a deformation model is learned without requiring (or otherwise utilizing) assumption of a skeletal system, which is appropriate for complex facial rigs.“ [0060] “The approximation may rely on CNNs, which generate deformation maps given input rig parameters.” [0065] “For each of the deformation maps, vertex offsets may be extracted by performing interpolation of the deformation map at each vertex position in texture coordinate space.” Examiner’s Note: Deformation maps modify vertex positions, are based on the target shape, and refine or approximate the final deformed mesh, and are therefore working as corrective displacement.)
Regarding Claim 10, Bailey in view of Le teaches the computer-implemented method of claim 9, wherein: the set of attributes comprises at least one of a bone point position, a soft tissue thickness, a bone normal, a skinning weight associated, or a set of corrective displacements (Bailey [0060] “deformation maps”); and the set of fitting parameters comprises at least one of an anatomical transformation or a set of blending coefficients associated with the set of corrective displacements. (Bailey [0060] “The approximation may rely on CNNs, which generate deformation maps given input rig parameters.” [0065] “For each of the deformation maps, vertex offsets may be extracted by performing interpolation of the deformation map at each vertex position in texture coordinate space.” Examiner’s Note: The rig parameters(fitting parameters) determine how the mesh is deformed; they act as blending coefficients that control the contribution of deformation influences on the final shape. Deformation maps modify vertex positions, are based on the target shape, and refine or approximate the final deformed mesh, and are therefore working as corrective displacement.)
Regarding Claim 11, Bailey in view of Le teaches one or more non-transitory computer readable media storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising: determining one or more ground truth positions (Bailey [0163] “facial landmarks“) of one or more points on a target shape associated with an object; (Bailey [0106] “deformed mesh output”); (Bailey [0163] “The facial landmarks are detected on the input image.”) generating, via execution of a set of neural networks, […] fitting parameters associated with […] the one or more points; (Bailey [0064] “For example, the rig parameters 302 may include inputs for controlling movement in the lip, eye brow, nose, etc.”) (Bailey [0163] “To animate the mesh, the movement of the detected landmarks in the recording was tracked and the IK model was used to estimate the rig parameters required to match the new landmark configuration.”) computing, via a shape model, (Bailey [0010] “a method for generating a mesh deformation of a facial model includes” Examiner Note: The shape model is the system that generates deformation maps via neural networks, extracts vertex offsets, and applies those offsets to achieve the output, as the system takes input parameters and outputs a shape, making it a shape model.) one or more predicted positions of the one or more points based on the set of fitting parameters(Bailey [0060] “According to at least one embodiment, a method of approximating utilizes artist-created texture coordinates U∈R.sup.n×2 of the facial mesh. The approximation may rely on CNNs, which generate deformation maps given input rig parameters.” [0065] “For each of the deformation maps, vertex offsets may be extracted by performing interpolation of the deformation map at each vertex position in texture coordinate space.” [0076] “With continued reference to FIG. 3, the coarse approximation computes the final vertex positions for the mesh by adding (e.g., at adder 310) the vertex offsets to the vertices of the neutral pose of the mesh”); training the set of neural networks based on one or more losses associated with the one or more predicted positions and the one or more ground truth positions (Bailey [0163] “The rig parameters are then passed to a disclosed approximation model to produce the deformed target mesh.” [0077] “Given the approximation model, a loss function is defined to find the optimal model parameters θ.sub.k. According to at least one embodiment, a loss function is proposed that penalizes both inaccuracies in approximated vertex positions, as well as inaccuracies in face normals on the mesh. Given a target mesh V, and approximated vertex offsets Δ, the loss function may be defined in Equation (1)” [0081] “Because the coarse approximation model works on separate mesh segments, the model could produce discontinuities across boundaries and/or seams between particular mesh segments. To address (e.g., minimize) this potential problem, the error function (e.g., the loss function of Equation (1)) may strongly penalize inaccurate face normals, to encourage smooth (or smoother) results along mesh segment boundaries. Penalizing normal errors also suppresses low-amplitude, high-frequency errors that may be visually disturbing or distracting.” Examiner Note: To penalize losses the predicted has to be compared to ground truth positions.); and generating, via execution of the trained set of neural networks, a three-dimensional (3D) model corresponding to the target shape (Bailey [0180] “a second plurality of deformation maps is generated by applying a second plurality of neural network-trained models” [0182] “a second plurality of vertex offsets is extracted based on the second plurality of deformation maps.” [0185] “The second plurality of vertex offsets is applied to the neutral mesh by applying the second plurality of vertex offsets to values of at most a subset of the vertices of the neutral mesh to generate the mesh deformation. The subset of the vertices of the neutral mesh may correspond to one or more regions of the neutral mesh exhibiting a level of approximation error that is above a particular threshold.”) but doesn’t explicitly disclose: […] per-point […] each of […].
Le teaches […] per-point […] each of […] (¶13, “Recently, learning-based techniques have been proposed to fit 3D primitives to a 3D point cloud. A recent work proposed a supervised learning-based framework called Supervised Primitive Fitting Network (SPFN) that learns a configuration of 3D primitives that represents a 3D point cloud. Instead of directly regressing representative primitive parameters such as dimensions, angles, or positions of 3D primitives, SPFN first predicts per-point features (association between each point and each primitive, primitive type, and surface normal), and a subsequent differentiable module computes representative primitive parameters by minimizing the fitting error. However, SPFN has limited accuracy and cannot efficiently handle high-resolution 3D point clouds. For example, the ability to process high-resolution 3D point clouds in a neural network with consumer GPUs can be limited by memory. For example, even with a relatively simple 3D point cloud processing architecture such as PointNet, memory limitations during training can limit input 3D point clouds to approximately 10 k points, whereas 3D point cloud scans may include 100 k to one million points. Simply downsampling a 3D point cloud results in the loss of fine detail. As a result, SPFN is not equipped to fit small 3D primitives that represent fine detail from high-resolution 3D point clouds.”).
Therefore it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to combine Bailey in view of Le as Supervised Primitive Fitting Network (SPFN) can robustly detect a varying number of geometric primitives at different scales from 3D point clouds without any manual per-input parameter tuning or user control.
Regarding Claim 12, Bailey in view of Le teaches the one or more non-transitory computer readable media of claim 11, wherein generating the set of fitting parameters comprises: generating, via execution of a first neural network included in the set of neural networks, one or more transformations associated with an anatomy of the object (Bailey [0010] “generating a first plurality of deformation maps by applying a first plurality of neural network-trained models; extracting a first plurality of vertex offsets based on the first plurality of deformation maps; and applying the first plurality of vertex offsets to a neutral mesh of the facial model to generate the mesh deformation of the facial model.” [0053] “In order to support inverse kinematics (IK) for a facial rig in real-time, an efficient and accurate inversion of the rig function may be necessary to compute character poses given a set of constraints.” Examiner’s Note: Changing character poses require a transformation of anatomy.); and generating, via execution of a second neural network included in the set of neural networks, a set of blending coefficients associated with a set of corrective displacements outputted by the shape model for the one or more points. (Bailey [0180] “a second plurality of deformation maps is generated by applying a second plurality of neural network-trained models” [0062] “To determine the position of the vertex, weighted sums of the pixel information in the deformation map (e.g., a weighted sum of the x-coordinate information of the two pixels, a weighted sum of the y-coordinate information of the two pixels, a weighted sum of the z-coordinate information of the two pixels) may be used to determine an offset for the vertex position in the mesh.” Examiner’s Note: Blending is the combination of multiple values using weights, which is done here with x, y, and z components combining using weighted sums.)
Regarding Claim 15, Bailey in view of Le teaches the one or more non-transitory computer readable media of claim 12, wherein: the 3D model comprises a deformation of a first face via the one or more transformations, (Bailey [Abstract] “and applying the first plurality of vertex offsets to a neutral mesh of the facial model to generate the mesh deformation of the facial model.”) and the one or more transformations are determined using a second face corresponding to the target shape (Bailey [Abstract] “generating a first plurality of deformation maps by applying a first plurality of neural network-trained models; extracting a first plurality of vertex offsets based on the first plurality of deformation maps” Examiner’s Note: The first face is the neutral mesh, second face is the face in the input image, which is used to generate the landmarks needed for rig parameters (Bailey [0163] which is then used for deformation maps.)
Regarding Claim 16, Bailey in view of Le teaches the one or more non-transitory computer readable media of claim 11, wherein the 3D model comprises a reconstruction of the target shape. (Bailey [Abstract] “and applying the first plurality of vertex offsets to a neutral mesh of the facial model to generate the mesh deformation of the facial model.” Examiner’s Note: The mesh deformation is a reconstruction of the targeted face.)
Regarding Claim 17, Bailey in view of Le teaches the one or more non-transitory computer readable media of claim 11, wherein the 3D model comprises an edit to an anatomy of the object. (Bailey [0013] “first rig parameter pose; providing a second plurality of vertices to a second network to produce a second rig parameter pose; and processing the rig parameter pose and the second rig parameter pose to produce a composite rig parameter pose” Examiner’s Note: Composite rig parameter pose is used as an edit to anatomy.)
Regarding Claim 18, Bailey in view of Le teaches the one or more non-transitory computer readable media of claim 11, wherein the shape model comprises an additional set of neural networks. (Bailey [0180] “a second plurality of deformation maps is generated by applying a second plurality of neural network-trained models” [0062] “To determine the position of the vertex, weighted sums of the pixel information in the deformation map (e.g., a weighted sum of the x-coordinate information of the two pixels, a weighted sum of the y-coordinate information of the two pixels, a weighted sum of the z-coordinate information of the two pixels) may be used to determine an offset for the vertex position in the mesh.”)
Regarding Claim 19, Bailey in view of Le teaches the one or more non-transitory computer readable media of claim 11, wherein the one or more ground truth positions comprise at least one of one or more 3D positions of the one or more points in a mesh associated with the target shape or one or more two-dimensional (2D) positions of the one or more points in an image of the target shape. (Bailey [0163] “The facial landmarks are detected on the input image.” Examiner’s Note: Ground truth points are 2D positions on an image of the target shape.)
Claim 20 is a system claim (Bailey [0128] illustrates a computer system) that recites similar limitations to Claim 11, and is therefore rejected under similar rationale.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to ROBERT J CRADDOCK whose telephone number is (571)270-7502. The examiner can normally be reached Monday - Friday 10:00 AM - 6 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, Devona E Faulk can be reached at 571-272-7515. 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.
/ROBERT J CRADDOCK/Primary Examiner, Art Unit 2618
Prosecution Insights