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 .
Information Disclosure Statement
The IDS(s) dated 12/13/2024 has/have been considered and placed in the application file.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
Claim(s) 1-20 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claim(s) 1 and 20 recite the limitation "a loss value that specifies a
difference between the one or more predicted transformations and the one or more respective reference transformations". There is insufficient antecedent basis for this limitation in the claim.
Claim(s) 2-19 depend either directly or indirectly from the rejection(s) of claim(s) 1, therefore they are also rejected.
Appropriate correction is required.
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 (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.
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) 1-9 and 14-20 is/are rejected under 35 U.S.C. 103 as obvious over Claessen et al (EP 3591616 A1, hereafter referred to as Claessen) and Qi et al (Qi, C. R., Su, H., Mo, K., & Guibas, L. J. (2017). Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 652-660), hereafter referred to as Qi), further in view of Aoki et al (Aoki, Y., Goforth, H., Srivatsan, R. A., & Lucey, S. (2019, June). Pointnetlk: Robust & efficient point cloud registration using pointnet. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7156-7165). IEEE., hereafter referred to as Aoki).
Claim 1
Regarding Claim 1, Claessen teaches “A computer-implemented method for training one or more neural networks to automatically generate coordinate systems used in digital oral care, the method comprising:
receiving, by one or more computer processors, a first digital 3D oral care representation of a patient's teeth (Claessen in ¶¶19-20, 52 discloses receiving voxel or surface mesh representations of a patient’s dental structure, including teeth);
receiving, by the one or more computer processors, one or more reference coordinate axes in proximity to one or more teeth in the first 3D oral care representation (Claessen in ¶¶30, 40, 65-66. 83-84 discloses supplying target data containing canonical coordinate values and defines canonical axes relative to identifiable dento-maxillofacial structures, including the dental arch); …
using, by the one or more computer processors, a first configuration of one or more neural networks that have been initially trained to generate a modified representation of the first digital 3D oral care representation (Claessen in ¶¶20, 73-80, 87-88 discloses a dental neural network predicts canonical coordinate information used to derive canonical axes);
automatically training, by the one or more computer processors, the second configuration, based on using the second configuration (Claessen in ¶¶40, 90-92 discloses training its canonical coordinate network by minimizing deviation between predicted canonical coordinates and target canonical coordinates and optimizing network parameters), wherein the training of the second configuration is modified by performing operations comprising:
predicting, by the second configuration, one or more predicted one or more directional vectors pertaining to the one or more coordinate axes (Claessen in ¶¶25, 78, 80, 84, 87 discloses predicting canonical coordinate voxel maps and deriving a local gradient 3D vector for each canonical axis, with the vector direction predicting the orientation of the corresponding axis);
computing, by the one or more computer processors, one or more predicted transformations from the one or more directional vectors (Claessen in ¶¶78-80, 87-88 discloses computing rotation parameters from the three predicted canonical axis directions and computing translation parameters from predicted canonical origin information); …
modifying at least one aspect of the one or more neural networks included in the second configuration based on the loss value (Claessen in ¶¶40, 90-92 discloses optimizing network parameters by minimizing prediction loss).”
Claessen does not explicitly teach all of “using, by the one or more computer processors, a first configuration of one or more neural networks that have been initially trained to generate a modified representation of the first digital 3D oral care representation; using, by the one or more computer processors, a second configuration of one or more neural networks that have been initially trained to predict information pertaining to one or more coordinate axes and wherein the second configuration receives as input the modified representation generated by the first configuration; automatically training, by the one or more computer processors, the second configuration, based on using the second configuration, wherein the training of the second configuration is modified by performing operations comprising: … determining, by the one or more computer processors, a loss value that specifies a difference between the one or more predicted transformations and the one or more respective reference transformations; and modifying at least one aspect of the one or more neural networks included in the second configuration based on the loss value.”
However, Qi teaches “using, by the one or more computer processors, a first configuration of one or more neural networks that have been initially trained to generate a modified representation of the first digital 3D oral care representation (Qi in pp. 654-655, Fig. 2, §4.2, “Joint Alignment Network” discloses a neural network input transform configuration that predicts an affine transformation matrix and applies the transformation to input point coordinates to generate an aligned point cloud representation);
using, by the one or more computer processors, a second configuration of one or more neural networks that have been initially trained to predict information pertaining to one or more coordinate axes and wherein the second configuration receives as input the modified representation generated by the first configuration; (Qi in Fig. 2, §4.2 discloses feeding a representation produced by an input alignment neural network to downstream neural network feature extraction and prediction layers. The proposed combination places Qi’s learned alignment configuration before Claessen’s canonical axis predictor)”
Qi is analogous art because Qi, like Claessen, is in the field of neural network processing of 3D spatial data and is reasonably pertinent to the problem of reducing pose variation before downstream prediction.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the input processing of Claessen with the learned input transformation configuration of Qi because Qi teaches that the learned transformation canonicalizes the input point set before downstream PointNet processing to improve the results (Qi, pp. 652, 654-655, Fig. 2, §4.2), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen would generate an aligned representation of the dental 3D data for Claessen’s canonical axis predictor, with a reasonable expectation of success because Claessen accepts dental surface mesh data comprising 3D points and Qi’s transformation network operates directly on the coordinates of input 3D points.
Claessen in view of Qi does not explicitly teach all of “automatically training, by the one or more computer processors, the second configuration, based on using the second configuration, wherein the training of the second configuration is modified by performing operations comprising: … determining, by the one or more computer processors, a loss value that specifies a difference between the one or more predicted transformations and the one or more respective reference transformations; and modifying at least one aspect of the one or more neural networks included in the second configuration based on the loss value.”
However, Aoki teaches “automatically training, by the one or more computer processors, the second configuration, based on using the second configuration (Aoki in pp. 4-5, §§3.3, 4.1, Fig. 2 discloses end to end training and fine tuning a PointNet-based transformation network using transformation error), wherein the training of the second configuration is modified by performing operations comprising: …
determining, by the one or more computer processors, a loss value that specifies a difference between the one or more predicted transformations and the one or more respective reference transformations (Aoki in p. 4, §3.3, Fig. 2 discloses training using a loss based on the difference between an estimated rigid transform and a ground truth transform); and
modifying at least one aspect of the one or more neural networks included in the second configuration based on the loss value (Aoki in pp. 4-5, §§3.3, 4.1 discloses transformation loss trains and fine tunes the PointNet-based network).”
Aoki is analogous art because Aoki, like Claessen and Qi, is in the field of neural network processing and alignment of 3D spatial data and is reasonably pertinent to the problem of training a network to minimize transformation error.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the loss based training of Claessen in view of Qi with the transformation level loss of Aoki because Aoki teaches that minimizing the difference between an estimated rigid transform and a ground truth transform provides transformation error for training a point cloud registration network (Aoki, p. 4, §3.3, Fig. 2), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen in view of Qi would train the second con figuration using the difference between predicted and reference transformations, with a reasonable expectation of success because Aoki’s transformation loss is differentiable and Aoki applies it to fine tune a PointNet-based network.
Thus, the claimed subject matter would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention.
Claim 2
Regarding Claim 2, Claessen and Qi, further in view of Aoki teaches “The computer-implemented method of claim 1, wherein the first digital specifies at least one of the patient's arches and further comprising data corresponding to one or more segmented teeth in at least one the patient's arches (Claessen in ¶¶30, 51-53 discloses a patient’s dental arch and a 3D surface mesh segmented into individual dental elements, including tooth crowns and surfaces belonging to gingiva).”
Claim 3
Regarding Claim 3, Claessen and Qi, further in view of Aoki teaches “The computer-implemented method of claim 1, wherein the at least one of the first configuration and the second configuration are initially trained using historical digital representations that includes one more coordinate axes (Claessen in ¶¶29, 40, 65-69 discloses initially training the dental neural network using multiple 3D dento-maxillofacial representations and associated target data comprising canonical coordinate values of a canonical coordinate system).”
Claim 4
Regarding Claim 4, Claessen and Qi, further in view of Aoki teaches “The computer-implemented method of claim 1, wherein the first representation comprises one or more mesh elements (Claessen in ¶¶19, 51 discloses a surface mesh comprising points or vertices connected by edges that define faces) and the method further comprises:
determining a mesh element feature vector for at least one of the mesh elements (Qi in p. 652 discloses representing each point by a vector of its three coordinates and computing normal and other local or global features);
providing, by the one or more computer processors, the mesh element feature vector as input to the first configuration (Qi in pp. 654-655, Fig. 2, §4.2 discloses providing the coordinates of input points to the T-Net input transform configuration); and
influencing the modified representation based on the mesh element feature vector (Qi in pp. 654-655, Fig. 2, §4.2 discloses predicting an affine transformation matrix from the input point data and applying the transformation to the input point coordinates to generate the aligned representation).”
Claim 5
Regarding Claim 5, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 1, wherein at least one neural network in any of the first configuration or the second configuration is trained, at least in part, using transfer learning (Aoki in p. 5, §4.1 discloses first training a standard PointNet classification network, initializing the PointNetLK feature extractor using that trained classification network, and fine tuning the feature extractor using the PointNetLK loss function).
Claim 6
Regarding Claim 6, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 1, wherein at least one neural network in any of the first configuration or the second configuration is used to train, at least in part, another neural network using transfer learning (Aoki in p. 5, §4.1 discloses first training a standard PointNet classification network, initializing the PointNetLK feature extractor using that trained classification network, and fine tuning the feature extractor using the PointNetLK loss function).
Claim 7
Regarding Claim 7, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 4, wherein the mesh element feature vector includes at least one spatial mesh element feature or at least one structural mesh element feature (Qi in p. 652 discloses representing each point by its spatial XYZ coordinates and computing normal and other local or global features).
Claim 8
Regarding Claim 8, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 7, wherein the mesh element comprises at least one of one or more vertices, one or more edges, one or more faces, one or more points of a point cloud, and one or more voxels of the first representation (Qi in pp. 651-652, Fig. 1 discloses a neural network that directly consumes points of a point cloud. Claessen in ¶¶19, 51 discloses 3D representations comprising voxels or surface meshes having points or vertices connected by edges that define faces).
Claim 9
Regarding Claim 9, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 8, wherein information pertaining to a vertex mesh element feature includes at least one or more of an XYZ position or a normal vector (Qi in p. 652 discloses representing each point by its XYZ coordinates and computing normals as point features).
Claim 14
Regarding Claim 14, Claessen and Qi, further in view of Aoki teaches The computer-implemented of claim 1, wherein one or more of the loss values that forms the basis of the modifying are selected from one or more of a binary cross entropy loss, mean squared error, an LI loss, and an L2 loss (Aoki in p. 4, §3.3 discloses expressing the difference between the estimated transformation and the ground truth transformation as mean square error between the estimated and ground truth twist parameters).
Claim 15
Regarding Claim 15, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 1, wherein one or more coordinate axes are automatically generated in real-time while the patient is present in the clinical environment (Claessen in ¶¶8, 10, 41 discloses automatically generating coordinate axes).
Claim 16
Regarding Claim 16, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 1, wherein the first digital representation describes at least one of one or more teeth, gingival tissues, and a dental or orthodontic appliance within the patient's mouth (Claessen in ¶¶4, 51-53 discloses intra-oral scan and surface mesh representations of a patient’s teeth and surrounding gingival surface).
Claim 17
Regarding Claim 17, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 1, wherein the predicted information includes at least one of one or more transformations or one or more vectors that are convertible into transformations (Claessen in ¶¶78-80, 87-88 discloses predicted axis direction vectors and canonical origin information that are used to calculate rotation and translation transformation parameters).
Claim 18
Regarding Claim 18, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 17, wherein at least one of two or more directional vectors or one or more positional vectors are generated by the second configuration (Claessen in ¶¶78-80, 87-88 discloses generating directional 3D vectors for the canonical axes directions and a translation vector identifying the position of the canonical origin in the coordinate prediction pipeline).
Claim 19
Regarding Claim 19, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 18, wherein the one or more computer processors use the directional vectors or positional vectors as input to generate at least one of three or more coordinate axes or the origin of the coordinate system (Claessen in ¶¶78-80, 87-88 discloses using the three predicted axis direction vectors to determine the canonical axes and using predicted positional information to determine canonical coordinate system origin).
Claim 20
Regarding Claim 20, Claessen teaches “A system comprising:
one or more computer processors;
non-transitory computer-readable storage having stored thereon first and second configurations of one or more neural networks and instructions that when executed by the one or more processors cause the one or more processors to:
receive a first digital 3D oral care representation of a patient's teeth (Claessen in ¶¶19-20, 52 discloses receiving voxel or surface mesh representations of a patient’s dental structure, including teeth);
receive one or more coordinate axes in proximity to one or more teeth in the first 3D oral care representation (Claessen in ¶¶30, 40, 65-66. 83-84 discloses supplying target data containing canonical coordinate values and defines canonical axes relative to identifiable dento-maxillofacial structures, including the dental arch); …
use the first configuration of one or more neural networks that have been initially trained to generate a modified representation of the first digital 3D oral care representation (Claessen in ¶¶20, 73-80, 87-88 discloses a dental neural network predicts canonical coordinate information used to derive canonical axes);
automatically train the second configuration, based on using the second configuration (Claessen in ¶¶40, 90-92 discloses training its canonical coordinate network by minimizing deviation between predicted canonical coordinates and target canonical coordinates and optimizing network parameters), wherein the training of the second configuration is modified by performing operations comprising:
predict, using the second configuration, one or more predicted transformations pertaining to the one or more coordinate axes (Claessen in ¶¶78-80, 87-88 discloses computing rotation parameters from the three predicted canonical axis directions and computing translation parameters from predicted canonical origin information); …
modify at least one aspect of the one or more neural networks included in the second configuration based on the loss value (Claessen in ¶¶40, 90-92 discloses optimizing network parameters by minimizing prediction loss).”
Claessen does not explicitly teach all of “use the first configuration of one or more neural networks that have been initially trained to generate a modified representation of the first digital 3D oral care representation; use the second configuration of one or more neural networks that have been initially trained to predict information pertaining to one or more coordinate axes and wherein the second configuration receives as input the modified representation generated by the first configuration; automatically train the second configuration, based on using the second configuration, wherein the training of the second configuration is modified by performing operations comprising: … determine a loss value that specifies a difference between the one or more predicted transformations and the one or more respective reference transformations; and modify at least one aspect of the one or more neural networks included in the second configuration based on the loss value.”
However, Qi teaches “use the first configuration of one or more neural networks that have been initially trained to generate a modified representation of the first digital 3D oral care representation (Qi in pp. 654-655, Fig. 2, §4.2, “Joint Alignment Network” discloses a neural network input transform configuration that predicts an affine transformation matrix and applies the transformation to input point coordinates to generate an aligned point cloud representation);
use the second configuration of one or more neural networks that have been initially trained to predict information pertaining to one or more coordinate axes and wherein the second configuration receives as input the modified representation generated by the first configuration; (Qi in Fig. 2, §4.2 discloses feeding a representation produced by an input alignment neural network to downstream neural network feature extraction and prediction layers. The proposed combination places Qi’s learned alignment configuration before Claessen’s canonical axis predictor)”
Qi is analogous art because Qi, like Claessen, is in the field of neural network processing of 3D spatial data and is reasonably pertinent to the problem of reducing pose variation before downstream prediction.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the input processing of Claessen with the learned input transformation configuration of Qi because Qi teaches that the learned transformation canonicalizes the input point set before downstream PointNet processing to improve the results (Qi, pp. 652, 654-655, Fig. 2, §4.2), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen would generate an aligned representation of the dental 3D data for Claessen’s canonical axis predictor, with a reasonable expectation of success because Claessen accepts dental surface mesh data comprising 3D points and Qi’s transformation network operates directly on the coordinates of input 3D points.
Claessen in view of Qi does not explicitly teach all of “automatically train the second configuration, based on using the second configuration, wherein the training of the second configuration is modified by performing operations comprising: … determine a loss value that specifies a difference between the one or more predicted transformations and the one or more respective reference transformations; and modify at least one aspect of the one or more neural networks included in the second configuration based on the loss value.”
However, Aoki teaches “automatically train the second configuration, based on using the second configuration (Aoki in pp. 4-5, §§3.3, 4.1, Fig. 2 discloses end to end training and fine tuning a PointNet-based transformation network using transformation error), wherein the training of the second configuration is modified by performing operations comprising: …
determine a loss value that specifies a difference between the one or more predicted transformations and the one or more respective reference transformations (Aoki in p. 4, §3.3, Fig. 2 discloses training using a loss based on the difference between an estimated rigid transform and a ground truth transform); and
modify at least one aspect of the one or more neural networks included in the second configuration based on the loss value (Aoki in pp. 4-5, §§3.3, 4.1 discloses transformation loss trains and fine tunes the PointNet-based network).”
Aoki is analogous art because Aoki, like Claessen and Qi, is in the field of neural network processing and alignment of 3D spatial data and is reasonably pertinent to the problem of training a network to minimize transformation error.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the loss based training of Claessen in view of Qi with the transformation level loss of Aoki because Aoki teaches that minimizing the difference between an estimated rigid transform and a ground truth transform provides transformation error for training a point cloud registration network (Aoki, p. 4, §3.3, Fig. 2), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen in view of Qi would train the second con figuration using the difference between predicted and reference transformations, with a reasonable expectation of success because Aoki’s transformation loss is differentiable and Aoki applies it to fine tune a PointNet-based network.
Thus, the claimed subject matter would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention.
Claim(s) 10 is/are rejected under 35 U.S.C. 103 as obvious over Claessen et al (EP 3591616 A1, hereafter referred to as Claessen), Qi et al (Qi, C. R., Su, H., Mo, K., & Guibas, L. J. (2017). Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 652-660), hereafter referred to as Qi), and Aoki et al (Aoki, Y., Goforth, H., Srivatsan, R. A., & Lucey, S. (2019, June). Pointnetlk: Robust & efficient point cloud registration using pointnet. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7156-7165). IEEE., hereafter referred to as Aoki), further in view of Touma et al (US 6167159 A, hereafter referred to as Touma).
Claim 10
Regarding Claim 10, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 9.
Claessen and Qi, further in view of Aoki does not explicitly teach all of wherein the normal vector is a weighted average of normal vectors of at least the connecting faces for the respective vertex.
However, Touma teaches wherein the normal vector is a weighted average of normal vectors of at least the connecting faces for the respective vertex (Touma in col. 27, lines 5-20, Fig. 18 discloses calculating a vertex prediction vector by adding the unit normal vectors of all triangles surrounding the vertex, which provides the direction of an equal weight average of the normals of the faces connected to the vertex).
Touma is analogous art because Touma, like Claessen, Qi, and Aoki, is in the field of processing geometric features of 3D mesh data and is reasonably pertinent to the problem of deriving vertex orientation information from the faces connected to a vertex.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the vertex feature processing of Claessen and Qi, further in view of Aoki with the connected face normals of Touma because Touma teaches that adding the unit normal vectors of all triangles surrounding the vertex provides a prediction vector that allows the normal list to encode the difference between the vertex normal and the prediction vector (col. 27, lines 5-20, Fig. 18), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen and Qi, further in view of Aoki would supply the direction of an equal weight average of the normals of the faces connected to the vertex as local surface orientation information, with a reasonable expectation of success because the dental surface mesh already provides the vertex to face connectivity and Touma’s operation uses ordinary vector addition of the connected face normals.
Thus, the claimed subject matter would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention.
Claim(s) 11 is/are rejected under 35 U.S.C. 103 as obvious over Claessen et al (EP 3591616 A1, hereafter referred to as Claessen), Qi et al (Qi, C. R., Su, H., Mo, K., & Guibas, L. J. (2017). Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 652-660), hereafter referred to as Qi), and Aoki et al (Aoki, Y., Goforth, H., Srivatsan, R. A., & Lucey, S. (2019, June). Pointnetlk: Robust & efficient point cloud registration using pointnet. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7156-7165). IEEE., hereafter referred to as Aoki), further in view of Verma et al (Verma, N., Boukhayma, A., Boyer, E., & Verbeek, J. (2021, December). Dual mesh convolutional networks for human shape correspondence. In 2021 International Conference on 3D Vision (3DV) (pp. 289-298). IEEE., hereafter referred to as Verma).
Claim 11
Regarding Claim 11, Claessen and Qi, further in view of Aoki teaches “The computer-implemented method of claim 8.”
Claessen and Qi, further in view of Aoki does not explicitly teach all of “wherein information pertaining to a face mesh element includes at least one or more of a XYZ position of a face centroid, face area, or a normal vector.”
However, Verma teaches “wherein information pertaining to a face mesh element includes at least one or more of a XYZ position of a face centroid, face area, or a normal vector (Verma in p. 4, §3.3, Fig. 3 discloses face level input features comprising the XYZ coordinates of the center of mass of the face, the surface area of the face, and a unit vector in the direction of the face normal).”
Verma is analogous art because Verma, like Claessen and Qi, further in view of Aoki, is in the field of neural network processing of 3D surface meshes and is reasonably pertinent to the problem of selecting face level geometric features for input to a mesh neural network.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the mesh element feature input of Claessen and Qi, further in view of Aoki with the face level geometric descriptors of Verma because Verma teaches that a face level feature comprising face center coordinates, face area, or a face normal vector provides a mesh neural network with information describing the face’s position, size, or orientation (p. 4, §3.3, Fig. 3), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen and Qi, further in view of Aoki would provide the network with information describing each dental mesh face’s position, size, or orientation, with a reasonable expectation of success because Verma supplies these descriptors directly as numeric inputs to a neural network operating on triangular surface meshes.
Thus, the claimed subject matter would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention.
Claim(s) 12 is/are rejected under 35 U.S.C. 103 as obvious over Claessen et al (EP 3591616 A1, hereafter referred to as Claessen), Qi et al (Qi, C. R., Su, H., Mo, K., & Guibas, L. J. (2017). Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 652-660), hereafter referred to as Qi), and Aoki et al (Aoki, Y., Goforth, H., Srivatsan, R. A., & Lucey, S. (2019, June). Pointnetlk: Robust & efficient point cloud registration using pointnet. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7156-7165). IEEE., hereafter referred to as Aoki), further in view of Hanocka et al (Hanocka, R., Hertz, A., Fish, N., Giryes, R., Fleishman, S., & Cohen-Or, D. (2019). Meshcnn: a network with an edge. ACM Transactions on Graphics (ToG), 38(4), 1-12., hereafter referred to as Hanocka).
Claim 12
Regarding Claim 12, Claessen and Qi, further in view of Aoki teaches The computer-implemented method of claim 8,
Claessen and Qi, further in view of Aoki does not explicitly teach all of wherein information pertaining to an edge mesh element include at least one or more of an XYZ position of an edge midpoint, an edge length, or a normal vector.
However, Hanocka teaches wherein information pertaining to an edge mesh element include at least one or more of an XYZ position of an edge midpoint, an edge length, or a normal vector.
Hanocka is analogous art because Hanocka, like Claessen and Qi, further in view of Aoki, is in the field of neural network processing of 3D surface meshes and is reasonably pertinent to the problem of selecting edge level geometric features for input to a mesh neural network.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the mesh element feature input of Claessen and Qi, further in view of Aoki with the edge midpoint coordinates of Hanocka because Hanocka teaches that an XYZ edge midpoint feature provides spatial position information as input to MeshCNN (p. 9, §5.4), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen and Qi, further in view of Aoki would provide the network with spatial position information for each edge of the dental surface mesh, with a reasonable expectation of success because Claessen’s surface meshes already comprise edges and Hanocka supplies the edge midpoint coordinates directly as numeric mesh network inputs.
Thus, the claimed subject matter would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention.
Claim(s) 13 is/are rejected under 35 U.S.C. 103 as obvious over Claessen et al (EP 3591616 A1, hereafter referred to as Claessen), Qi et al (Qi, C. R., Su, H., Mo, K., & Guibas, L. J. (2017). Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 652-660), hereafter referred to as Qi), Aoki et al (Aoki, Y., Goforth, H., Srivatsan, R. A., & Lucey, S. (2019, June). Pointnetlk: Robust & efficient point cloud registration using pointnet. In 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (pp. 7156-7165). IEEE., hereafter referred to as Aoki), and Hanocka et al (Hanocka, R., Hertz, A., Fish, N., Giryes, R., Fleishman, S., & Cohen-Or, D. (2019). Meshcnn: a network with an edge. ACM Transactions on Graphics (ToG), 38(4), 1-12., hereafter referred to as Hanocka), further in view of Biermann et al (US 20020191863 A1, hereafter referred to as Biermann).
Claim 13
Regarding Claim 13, Claessen, Qi, and Aoki, further in view of Hanocka teaches The computer-implemented method of claim 12.
Claessen, Qi, and Aoki, further in view of Hanocka does not explicitly teach all of wherein the normal vector is an average of the normal vectors of at least two vertices.
However, Biermann teaches wherein the normal vector is an average of the normal vectors of at least two vertices (Biermann in ¶88 discloses computing a normal at a point on an edge as the average of the normal vectors at the two vertices defining the edge).
Biermann is analogous art because Biermann, like Claessen, Qi, and Aoki, further in view of Hanocka, is in the field of processing local geometry on 3D surface meshes and is reasonably pertinent to the problem of deriving a surface normal vector at a point along a mesh edge.
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify the edge feature processing of Claessen, Qi, and Aoki, further in view of Hanocka with the endpoint normal interpolation of Biermann because Bierman teaches that averaging the normal vectors at two vertices defining an edge provides a normal at a point on the edge for approximating a smoothly changing surface normal (¶¶84-88), and one of ordinary skill would have recognized that incorporating this feature into the method of Claessen, Qi, and Aoki, further in view of Hanocka would generate a normal vector associated with the edge midpoint from two endpoint vertex normals, with a reasonable expectation of success because the endpoint normals and edge connectivity are existing mesh data and Biermann’s averaging operation produces the edge associated normal.
Thus, the claimed subject matter would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JUSTIN P CASCAIS whose telephone number is (703) 756-5576. The examiner can normally be reached Monday-Friday 8:00-4:00.
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, Mr. O'Neal Mistry can be reached on (313) 446-4912. 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.
/J.P.C./Examiner, Art Unit 2674
/ONEAL R MISTRY/Supervisory Patent Examiner, Art Unit 2674
Date: ***
Prosecution Insights