Mesh optimization using novel segmentation

Abstract

A method and apparatus for rendering a mesh from one or more virtual camera views of an object, the one or more virtual camera views being projections of the mesh onto multiple 2D planes, inputting the one or more virtual camera views into a fully convolutional model for real-time segmentation, generating one or more 2D pixel-wise masks based on the fully convolutional model, the 2D pixel-wise masks being associated with one or more segmented elements, performing remeshing and mesh optimization using the one or more 2D pixel-wise masks, and generating an output mesh based on the remeshing and mesh optimization.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Patent Application No. 63 / 398,672, filed August 17, 2022, and U.S. Patent Application No. 18 / 315,125, filed May 10, 2023, the disclosures of which are incorporated herein by reference in their entireties.

[0002] This disclosure relates to mesh and video-based mesh compression techniques. More specifically, a system for achieving high-quality volume segmentation masks and methods for mesh optimization and remeshing are disclosed. [Background technology]

[0003] Advances in 3D capture, modeling, and rendering are facilitating the ubiquitous presence of 3D content across several platforms and devices. Today, it is possible to film a baby's first steps on one continent, while the baby's grandparents on another continent watch (and even interact with) this, enjoying a fully immersive experience with the child. Nevertheless, to achieve this sense of realism, models are becoming ever more sophisticated, and a significant amount of data is tied to the creation and consumption of these models. 3D meshes are widely used to represent such immersive content.

[0004] Dynamic mesh sequences can require large amounts of data because they can consist of a significant amount of information that changes over time. Therefore, efficient compression techniques are needed to store and transmit such content. The mesh compression standards IC, MESHGRID, and FAMC were previously developed by MPEG to address dynamic meshes with constant connectivity and time-varying geometry and vertex attributes. However, these standards do not take time-varying attribute maps and connectivity information into account. Digital content creation (DCC) tools typically generate such dynamic meshes. Correspondingly, it is difficult for volumetric acquisition techniques to generate constant connectivity dynamic meshes, especially under real-time constraints. This type of content is not supported by existing standards. Uncompressed or raw mesh and texture data for dynamic human subjects and objects can result in very large file sizes. Current mesh optimization techniques rely too heavily on manual processing, preventing real-time capture into a transmission pipeline. Summary of the Invention [Means for solving the problem]

[0005] The following presents a simplified summary of one or more embodiments of the present disclosure in order to provide a basic understanding of such embodiments. This summary is not an extensive overview of all contemplated embodiments, and is not intended to identify key or critical elements of all embodiments or to delineate the scope of any or all embodiments. Its sole purpose is to present some concepts of one or more embodiments of the present disclosure in a simplified form as a prelude to the more detailed description that is presented later.

[0006] The present disclosure provides a system for achieving high quality volume segmentation masks and methods for mesh optimization and remeshing.

[0007] According to some embodiments, a method implemented by at least one processor is provided that may include rendering a mesh from one or more virtual camera views of an object, where the one or more virtual camera views are projections of the mesh onto multiple 2D planes. The method may further include inputting the virtual camera views into a fully convolutional model for real-time segmentation. The method may further include generating one or more 2D pixel-wise masks based on the fully convolutional model, where the 2D pixel-wise masks are associated with one or more segmented elements. The method may further include performing remeshing and mesh optimization using the one or more 2D pixel-wise masks. The method may further include generating an output mesh based on the remeshing and mesh optimization.

[0008] According to other aspects of one or more embodiments, an apparatus and a non-transitory computer-readable medium consistent with the method are also provided.

[0009] Additional embodiments will be set forth in the description that follows, and in part will be obvious from the description, and / or may be learned by practice of presented embodiments of the present disclosure.

[0010] Further features, nature and various advantages of the disclosed subject matter will become more apparent from the following detailed description and accompanying drawings. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is a schematic diagram of a simplified block diagram of a communication system, according to some embodiments. [Figure 2] FIG. 1 is a schematic diagram of a simplified block diagram of a streaming system, according to some embodiments. [Figure 3] FIG. 1 is a diagram of multiple virtual cameras used to infer a pixel-wise mask, according to some embodiments. [Figure 4] FIG. 1 illustrates an example of instance segmentation, according to some embodiments. [Figure 5] FIG. 10 illustrates an example of an optimized remeshing segment, according to some embodiments. [Figure 6] 1 is an operational flowchart illustrating steps performed by a program for mesh optimization and remeshing, according to some embodiments. [Figure 7] FIG. 1 is a diagram of a computer system suitable for implementing embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0012] The following detailed description of the exemplary embodiments refers to the accompanying drawings, in which the same reference numbers in different drawings may identify the same or similar elements.

[0013] In this disclosure, several methods are proposed for achieving high-quality volume segmentation masks and mesh optimization and remeshing. These methods can be applied individually or in any combination. Furthermore, each of the methods (or embodiments), encoders, and decoders may be implemented by processing circuitry (e.g., one or more processors or one or more integrated circuits). In one example, the one or more processors execute a program stored on a non-transitory computer-readable medium.

[0014] 1-2, one embodiment of the present disclosure for implementing the encoding and decoding structures of the present disclosure will be described.

[0015] 1 shows a simplified block diagram of a communication system 100 according to one embodiment of the present disclosure. The system 100 may include at least two terminals 110, 120 interconnected via a network 150. In the case of unidirectional data transmission, the first terminal 110 may code video data, which may include mesh data, at a local location for transmission to the other terminal 120 via the network 150. The second terminal 120 may receive the coded video data of the other terminal from the network 150, decode the coded data, and display the recovered video data. The unidirectional data transmission may be used in media serving applications, etc.

[0016] 1 shows a second pair of terminals 130, 140 provided to support bidirectional transmission of coded video, such as may occur during a video conference. For bidirectional transmission of data, each terminal 130, 140 can code video data captured at a local location for transmission to the other terminal over network 150. Each terminal 130, 140 can also receive coded video data transmitted by the other terminal, decode the coded data, and display the recovered video data on a local display device.

[0017] In FIG. 1 , terminals 110-140 may be, for example, servers, personal computers, and smartphones, and / or any other type of terminal. For example, terminals 110-140 may be laptop computers, tablet computers, media players, and / or dedicated videoconferencing equipment. Network 150 represents any number of networks that convey coded video data between terminals 110-140, including, for example, wired and / or wireless communication networks. Communication network 150 may exchange data over circuit-switched and / or packet-switched channels. Exemplary networks include telecommunications networks, local area networks, wide area networks, and / or the Internet. For purposes of this discussion, the architecture and topology of network 150 may not be important to the operation of the present disclosure, unless described herein below.

[0018] 2 illustrates the arrangement of a video encoder and a video decoder in a streaming environment as an example of an application for the disclosed subject matter. The disclosed subject matter may also be used in other video-enabled applications, including, for example, video conferencing, digital TV, storage of compressed video on digital media including CDs, DVDs, memory sticks, etc.

[0019] 2, the streaming system 200 may include a capture subsystem 213 that includes a video source 201 and an encoder 203. The streaming system 200 may further include at least one streaming server 205 and / or at least one streaming client 206.

[0020] The video source 201 may create a stream 202 including, for example, a 3D mesh and metadata associated with the 3D mesh. The video source 201 may include, for example, a 3D sensor (e.g., a depth sensor) or 3D imaging technology (e.g., a digital camera) and a computing device configured to generate the 3D mesh using data received from the 3D sensor or 3D imaging technology. The sample stream 202 may have a high amount of data compared to an encoded video bitstream and may be processed by an encoder 203 coupled to the video source 201. The encoder 203 may include hardware, software, or a combination thereof that enables or implements aspects of the disclosed subject matter, as described in more detail below. The encoder 203 may also generate an encoded video bitstream 204. The encoded video bitstream 204 may have a low amount of data compared to the uncompressed stream 202 and may be stored on a streaming server 205 for later use. One or more streaming clients 206 can access the streaming server 205 and retrieve a video bitstream 209 , which may be a copy of the encoded video bitstream 204 .

[0021] The streaming client 206 may include a video decoder 210 and a display 212. The video decoder 210 may, for example, decode a video bitstream 209 that is a copy of the input encoded video bitstream 204 and create an output video sample stream 211 that can be rendered on the display 212 or another rendering device (not shown). In some streaming systems, the video bitstreams 204, 209 may be encoded according to a particular video coding / compression standard.

[0022] A mesh generally refers to a number of polygons that describe the surface of a volumetric object. Information about its vertices in 3D space and how the vertices are connected can define each polygon, referred to as connectivity information. Optionally, vertex attributes, such as color and normals, can be associated with mesh vertices. Attributes can also be associated with the surface of a mesh by utilizing mapping information that parameterizes the mesh with a 2D attribute map. Such mappings are sometimes defined using a set of parametric coordinates, called UV coordinates or texture coordinates, associated with mesh vertices. 2D attribute maps can be used to store high-resolution attribute information, such as texture, normals, and displacements. Such information can be used for various purposes, such as texture mapping, shading, and mesh reconstruction.

[0023] In some embodiments, the mesh may be rendered from one or more virtual camera views, where the views may be considered as projections of the 3D mesh data onto a 2D plane. Rendering can refer to a process performed on 2D data to make the 2D data appear volumetric and three-dimensional. These views are fed into a fully convolutional model or other system for real-time segmentation. The resulting 2D pixel-wise mask associated with the segmented elements can then be used as additional input to enhance downstream remeshing and mesh optimization. The remeshing process can include removing overlapping polygons when rendering a new mesh. Mesh optimization can refer to the process of lowering the bandwidth required to render a mesh.

[0024] A pixel-wise mask generally refers to a semantically segmented set of pixels that assigns pixel-wise correlations of objects in a 2D frame. For example, if there are four penguins in a scene, each penguin is assigned a pixel-wise mask that represents all pixels. In this case, there will be four instance segmented pixel-wise masks for penguins 1-4.

[0025] The pixel-wise mask can also be assumed to carry associated metadata correlating the virtual camera used to render the scene to the mask, such as camera pose, field of view, virtual sensor size, resolution, aspect ratio, etc. As seen in Figure 3, multiple virtual cameras 301-304 can be configured to render multiple views used to infer the pixel-wise mask of object 300.

[0026] 4 shows an example of multiple pixel-wise masks rendered by a virtual camera, according to some embodiments. The multiple pixel-wise masks and their associated metadata can be used to generate an output mesh that is enhanced in multiple ways. These enhancements may include, but are not limited to: Preservation of facial features and frequent objects of interest, as well as details of known human visual saliency features · Removal of redundant vertices and points, remapping of index and vertex buffers. Integration with mesh shader pipeline for vertex reuse and culling.

[0027] 5 shows an example of optimized remeshing according to some embodiments. The dashed lines represent polygons that may include pixel-wise masks. The remeshing process can include optimizing the mesh by removing overlapping polygons and reducing bandwidth.

[0028] 6 is a flowchart of an example process 600 for achieving a high-quality volume segmentation mask using mesh optimization and remeshing. In some embodiments, one or more process blocks of FIG. 6 may be performed by any of the elements described above.

[0029] As shown in FIG. 6, process 600 can include rendering a mesh from one or more virtual camera views of an object, where the one or more virtual camera views are projections of the mesh onto multiple 2D planes (block 610).

[0030] As further shown in FIG. 6, process 600 may include inputting the virtual camera view into a fully convolutional model for real-time segmentation (block 620).

[0031] As further shown in FIG. 6, process 600 may include generating one or more 2D pixel-wise masks based on the fully convolutional model, the 2D pixel-wise masks being associated with one or more segmented elements (block 630).

[0032] As further shown in FIG. 6, the process 600 may include performing remeshing and mesh optimization using one or more 2D pixel-wise masks (block 640).

[0033] As further shown in FIG. 6, the process 600 may include generating an output mesh based on the remeshing and mesh optimization (block 650).

[0034] Although Figure 6 illustrates example blocks of process 600, in some implementations, process 600 may include additional, fewer, different, or differently arranged blocks relative to those illustrated in Figure 6. Additionally or alternatively, two or more blocks of process 600 may be performed in parallel.

[0035] Below are selected embodiments of the proposed method and system.

[0036] In some embodiments, the masks are created from multiple angles using a trained convolutional neural network.

[0037] In some embodiments, the mask is created from multiple angles using non-AI based logic.

[0038] In some embodiments, separate masks can be generated to further segment facial elements, arms, legs, etc. separately.

[0039] In some embodiments, the pose and orientation of the virtual camera is set based on knowledge of previously trained data, such as a human 3D portrait photography dataset.

[0040] In some embodiments, the pixel-wise mask is used as input to a video-based mesh compression system.

[0041] In some embodiments, a pixel-wise mask may be used to subdivide the mesh into multiple segments, and each segment may be packed into a different geometry map and then coded into a different sub-stream.

[0042] In some embodiments, a pixel-wise mask may be used to determine the coding parameters for each segment, e.g., human face segments may use a smaller quantization parameter while being coded by a video codec.

[0043] In some embodiments, a pixel-wise mask may be used to perform the remeshing operation. For example, in Figure 5, different remeshing operations or different quantizations may be applied to different segments.

[0044] In some embodiments, the pixel-wise mask may be used as input to a scene graph processor for streaming scene graph elements to multiple decoders, intermediate software, and renderers.

[0045] The techniques described above can be implemented as computer software using computer-readable instructions and physically stored on one or more computer-readable media. For example, Figure 7 illustrates a computer system 900 suitable for implementing certain embodiments of the present disclosure.

[0046] The computer software may be coded using any suitable machine code or computer language that can undergo mechanisms such as assembly, compilation, linking, etc. to produce code containing instructions that can be executed by a computer central processing unit (CPU), graphics processing unit (GPU), etc. directly, or via interpretation, microcode execution, etc.

[0047] The instructions may be executed on various types of computers or components thereof, including, for example, personal computers, tablet computers, servers, smartphones, gaming consoles, Internet of Things devices, and the like.

[0048] 7 for computer system 900 are examples and are not intended to suggest any limitation as to the scope of use or functionality of the computer software implementing embodiments of the present disclosure. The arrangement of components should not be interpreted as having any dependency or requirement regarding any one or combination of components illustrated in the non-limiting embodiment of computer system 900.

[0049] The computer system 900 may include certain human interface input devices. Such human interface input devices may respond to input by one or more human users, for example, via tactile input (e.g., keystrokes, swipes, data glove movements), audio input (e.g., voice, clapping), visual input (e.g., gestures), or olfactory input (not shown). The human interface devices may also be used to capture certain media not necessarily directly associated with conscious human input, such as audio (e.g., voice, music, ambient sounds), images (e.g., scanned images, photographic images, still image captured from a camera), and video (e.g., two-dimensional video, three-dimensional video including stereoscopic video).

[0050] The input human interface devices may include one or more of a keyboard 901, a mouse 902, a trackpad 903, a touchscreen 910, a data glove, a joystick 905, a microphone 906, a scanner 907, and a camera 908 (only one of each is shown).

[0051] The computer system 900 may also include certain human interface output devices. Such human interface output devices may stimulate one or more of the human user's senses, for example, through tactile output, sound, light, and smell / taste. Such human interface output devices may include haptic output devices (e.g., haptic feedback via a touchscreen 910, data gloves, or joystick 905, although haptic feedback devices that do not function as input devices may also be present). For example, such devices may be audio output devices (such as speakers 909, headphones (not shown)), visual output devices (such as screens 910 including CRT screens, LCD screens, plasma screens, and OLED screens, each with or without touchscreen input capability, each with or without haptic feedback capability, some of which may be capable of outputting two-dimensional visual output or output in more than three dimensions through means such as stereoscopic output, virtual reality glasses (not shown), holographic displays, and smoke tanks (not shown)), and printers (not shown).

[0052] The computer system 900 may also include human-accessible storage devices and their associated media, such as optical media including CD / DVD ROM / RW 920 with CD / DVD or similar media 921, thumb drives 922, removable hard drives or solid state drives 923, legacy magnetic media such as tape and floppy disks (not shown), and specialized ROM / ASIC / PLD-based devices (not shown) such as security dongles.

[0053] Those skilled in the art should also understand that the term "computer-readable medium" as used in connection with the subject matter of this disclosure does not encompass transmission media, carrier waves, or other transitory signals.

[0054] The computer system 900 may also include interfaces to one or more communication networks. The networks may be, for example, wireless, wired, or optical. The networks may also be local, wide area, metropolitan, vehicular, industrial, real-time, delay-tolerant, etc. Examples of networks include local area networks such as Ethernet, WLAN, etc.; cellular networks including GSM, 3G, 4G, 5G, LTE, etc.; TV wired or wireless wide area digital networks including cable TV, satellite TV, and terrestrial broadcast TV; and vehicular and industrial networks including CANBus. Certain networks generally require an external network interface adapter attached to a particular general-purpose data port or peripheral bus 949 (e.g., a USB port of the computer system 900, etc.), while other networks are generally integrated into the core of the computer system 900 by attachment to the system bus as described below (e.g., an Ethernet interface to a PC computer system, or a cellular network interface to a smartphone computer system). Using any of these networks, the computer system 900 can communicate with other entities. Such communications may be one-way receive only (e.g., broadcast TV), one-way transmit only (e.g., a CANbus to a particular CANbus device), or two-way, for example, to other computer systems using local-area or wide-area digital networks. Such communications may include communications to a cloud computing environment 955. Certain protocols and protocol stacks may be used with each of these networks and network interfaces, as described above.

[0055] The aforementioned human interface devices, human-accessible storage devices, and network interface 954 may be attached to core 940 of computer system 900 .

[0056] The core 940 may include one or more central processing units (CPUs) 941, graphics processing units (GPUs) 942, dedicated programmable processing units in the form of field programmable gate arrays (FPGAs) 943, hardware accelerators 944 for specific tasks, etc. These devices may be connected via a system bus 948, along with read-only memory (ROM) 945, random access memory 946, and internal mass storage 947, such as a non-user-accessible internal hard drive or SSD. In some computer systems, the system bus 948 may be accessible in the form of one or more physical plugs to allow expansion with additional CPUs, GPUs, etc. Peripheral devices may be attached directly to the core's system bus 948 or via a peripheral bus 949. Architectures for peripheral buses include PCI, USB, etc. A graphics adapter 950 may also be included in the core 940.

[0057] The CPU 941, GPU 942, FPGA 943, and accelerator 944 may combine to execute specific instructions that may constitute the aforementioned computer code, which may be stored in ROM 945 or RAM 946. Persistent data may be stored in, for example, internal mass storage 947, while temporary data may also be stored in RAM 946. Rapid storage and retrieval in any of the memory devices may be enabled through the use of cache memory, which may be closely associated with one or more of the CPU 941, GPU 942, mass storage 947, ROM 945, RAM 946, etc.

[0058] The computer-readable medium may bear computer code for performing various computer-implemented operations. The medium and computer code may be those specially designed and constructed for the purposes of the present disclosure, or they may be of the kind well known and available to those skilled in the computer software arts.

[0059] By way of example and not limitation, a computer system having the architecture of computer system 900, and specifically core 940, may provide functionality as a result of a processor (including a CPU, GPU, FPGA, accelerator, etc.) executing software embodied in one or more tangible computer-readable media. Such computer-readable media may be user-accessible mass storage as introduced above, as well as media associated with specific storage of core 940 that is non-transitory in nature, such as core-internal mass storage 947 or ROM 945. Software implementing various embodiments of the present disclosure may be stored on such devices and executed by core 940. The computer-readable media may include one or more memory devices or chips, depending on particular needs. The software may cause core 940, and specifically the processor therein (including a CPU, GPU, FPGA, etc.), to perform particular processes or particular portions of particular processes described herein, including defining data structures stored in RAM 946 and modifying such data structures according to the software-defined processes. Additionally or alternatively, a computer system may provide functionality as a result of logic hardwired or otherwise embodied in circuitry (e.g., accelerator 944), which may operate in place of or in conjunction with software to perform particular processes or portions of particular processes described herein. References to software may also encompass logic, where appropriate, and vice versa. References to computer-readable media may encompass circuitry (such as an integrated circuit (IC)) that stores software for execution, circuitry that embodies logic for execution, or both, where appropriate. The present disclosure encompasses any suitable combination of hardware and software.

[0060] While this disclosure describes several non-limiting embodiments, there are alterations, substitutions, and various substitute equivalents that fall within the scope of this disclosure. It will thus be appreciated that those skilled in the art will be able to devise numerous systems and methods that, although not explicitly shown or described herein, embody the principles of the present disclosure and are therefore within the spirit and scope of the present disclosure. [Explanation of symbols]

[0061] 100 Communication Systems 110 First Terminal 120 Second Terminal 130 terminals 140 terminals 150 Communication Network 200 Streaming System 201 Video Sources 202 Sample Stream 203 Encoder 204 encoded video bitstream 205 Streaming Server 206 Streaming Client 209 Video Bitstream 210 Video Decoder 211 video sample streams 212 Display 213 Capture Subsystem 300 objects 301 Virtual Camera 302 Virtual Camera 303 Virtual Camera 304 Virtual Camera 600 processes 900 Computer Systems 901 Keyboard 902 Mouse 903 Trackpad 905 Joystick 906 Microphone 907 Scanner 908 Camera 909 Speaker 910 Touchscreen 920 CD / DVD ROM / RW 921 CD / DVD or similar media 922 thumb drive 923 Removable Hard Drive or Solid State Drive 940 cores 941 Central Processing Unit (CPU) 942 Graphics Processing Unit (GPU) 943 Field Programmable Gate Area (FPGA) 944 Hardware Accelerator 945 Read-Only Memory (ROM) 946 Random Access Memory (RAM) 947 Internal Mass Storage 948 System Bus 949 Peripheral Bus 950 graphics adapter 954 network interface 955 Cloud Computing Environment

Claims

1. 1. A method implemented by at least one processor, comprising: Rendering a mesh from one or more virtual camera views of an object, the one or more virtual camera views being projections of the mesh onto a plurality of two-dimensional (2D) planes; inputting the one or more virtual camera views into a fully convolutional model for real-time segmentation; generating one or more 2D pixel-wise masks based on the fully convolutional model, the 2D pixel-wise masks being associated with one or more segmented elements; performing remeshing and mesh optimization using said one or more 2D pixel-wise masks; generating an output mesh based on the remeshing and mesh optimization; A method comprising:

2. 10. The method of claim 1, wherein the one or more 2D pixel-wise masks are generated from multiple angles using a trained convolutional neural network.

3. The method of claim 1 , wherein the one or more 2D pixel-wise masks are generated from multiple angles using logic that is not based on artificial intelligence.

4. The method of claim 1 , further comprising generating one or more separate 2D pixel-wise masks to further segment one or more elements of the object.

5. The method of claim 1 , further comprising setting the pose and orientation of the virtual camera based on knowledge of previously trained data.

6. The method of claim 1 , wherein the one or more 2D pixel-wise masks are used as input to a video-based mesh compression system.

7. subdividing the mesh into a plurality of segments using the one or more 2D pixel-wise masks; packing each of the plurality of segments into a different geometry map; coding the various geometry maps into multiple sub-streams; The method of claim 1 further comprising:

8. The method of claim 7 , wherein the one or more 2D pixel-wise masks are used to determine one or more coding parameters for each of the plurality of segments.

9. The method of claim 1 , wherein the one or more 2D pixel-wise masks are used as input to a scene graph processor for streaming multiple scene graph elements to multiple decoders, intermediate software, and renderers.

10. at least one memory configured to store program code; at least one processor configured to read said program code and to operate as instructed by said program code, said program code comprising: rendering code configured to cause the at least one processor to render a mesh from one or more virtual camera views of an object, the one or more virtual camera views being projections of the mesh onto a plurality of two-dimensional (2D) planes; input code configured to cause the at least one processor to input the one or more virtual camera views into a fully convolutional model for real-time segmentation; first generation code configured to cause the at least one processor to generate one or more 2D pixel-wise masks based on the fully convolutional model, the 2D pixel-wise masks being associated with one or more segmented elements; and remeshing code configured to cause the at least one processor to perform remeshing and mesh optimization using the one or more 2D pixel-wise masks; a second generation code configured to cause the at least one processor to generate an output mesh based on the remeshing and mesh optimization; and At least one processor including An apparatus comprising:

11. 11. The apparatus of claim 10, wherein the one or more 2D pixel-wise masks are generated from multiple angles using a trained convolutional neural network.

12. The apparatus of claim 10 , wherein the one or more 2D pixel-wise masks are generated from multiple angles using logic that is not based on artificial intelligence.

13. 11. The apparatus of claim 10, wherein the program code further includes third generation code configured to cause the at least one processor to generate one or more separate 2D pixel-wise masks to further segment one or more elements of the object.

14. 11. The apparatus of claim 10, wherein the program code further comprises setting code configured to cause the at least one processor to set a pose and orientation of a virtual camera based on knowledge of previous training data.

15. The apparatus of claim 10 , wherein the one or more 2D pixel-wise masks are used as input to a video-based mesh compression system.

16. The program code subdivision code configured to cause the at least one processor to subdivide the mesh into a plurality of segments using the one or more 2D pixel-wise masks; packing code configured to cause the at least one processor to pack each of the plurality of segments into a different geometry map; substream code configured to cause the at least one processor to code the various geometry maps into a plurality of substreams; The apparatus of claim 10 further comprising:

17. 17. The apparatus of claim 16, wherein the one or more 2D pixel-wise masks are used to determine one or more coding parameters for each of the plurality of segments.

18. 11. The apparatus of claim 10, wherein the one or more 2D pixel-wise masks are used as input to a scene graph processor for streaming multiple scene graph elements to multiple decoders, intermediate software, and renderers.

19. A non-transitory computer-readable storage medium storing instructions that, when executed by at least one processor, cause the at least one processor to: rendering a mesh from one or more virtual camera views of an object, the one or more virtual camera views being projections of the mesh onto a plurality of two-dimensional (2D) planes; inputting the one or more virtual camera views into a fully convolutional model for real-time segmentation; generating one or more 2D pixel-wise masks based on the fully convolutional model, the 2D pixel-wise masks being associated with one or more segmented elements; performing remeshing and mesh optimization using said one or more 2D pixel-wise masks; generating an output mesh based on the remeshing and mesh optimization; A non-transitory computer-readable storage medium.

20. 20. The non-transitory computer-readable storage medium of claim 19, wherein the one or more 2D pixel-wise masks are generated from multiple angles using a trained convolutional neural network.

Images