Method and an apparatus for reconstructing an occupancy map of a point cloud frame
At least one embodiment relates to a method and an apparatus for reconstructing an occupancy map comprising occupancy data of a volumetric content, wherein reconstructing the occupancy map comprises: —decoding the occupancy map at a first resolution, —determining a scale factor as a function of the first resolution, —upscaling the occupancy map by the scale factor, using a neural network.
1 . A method, comprising:
signaling, in a bitstream, data representative of a volumetric content, the volumetric content being represented as a set of projections onto one or more atlases, the one or more atlases being video-based encoded, the one or more atlases comprising at least one attribute frame, one geometry frame and one occupancy map, and
signaling, in the bitstream, an information indicating a method among at least two methods for upscaling at least one of the one or more atlases when reconstructing the volumetric content,
wherein the at least one of the one or more atlases is an occupancy map, and
wherein one of the at least two methods is a neural network-based upscaling.
2 . The method of claim 1 , wherein the information further comprises at least one of an indicator representative of a neural network model, a size of block used as input of a neural network or a number of iterations according to which upscaling of the occupancy map is iterated.
3 . An apparatus, comprising one or more processors, wherein the one or more processors are configured to:
signal, in a bitstream, data representative of a volumetric content, the volumetric content being represented as a set of projections onto one or more atlases, the one or more atlases being video-based encoded, the one or more atlases comprising at least one attribute frame, one geometry frame and one occupancy map, and
signal, in the bitstream, an information indicating a method among at least two methods for upscaling at least one of the one or more atlases when reconstructing the volumetric content,
wherein the at least one of the one or more atlases is an occupancy map, and
wherein one of the at least two methods is a neural network-based upscaling.
4 . The apparatus of claim 3 , wherein the information further comprises at least one of an indicator representative of a neural network model, a size of block used as input of a neural network or a number of iterations according to which upscaling of the occupancy map is iterated.
5 . A method, comprising reconstructing a volumetric content from a bitstream, the volumetric content being represented as a set of projections onto one or more atlases, the one or more atlases being video-based encoded, the one or more atlases comprising at least one attribute frame, one geometry frame and one occupancy map, wherein reconstructing the volumetric content comprises:
decoding an information indicating a method among at least two methods for upscaling at least one of the one or more atlases, wherein one of the at least two methods is a neural network-based upscaling,
wherein the at least one of the one or more atlases is an occupancy map,
decoding the occupancy map from the bitstream, and
upscaling the occupancy map.
6 . The method of claim 5 , wherein the information indicates use of a neural network for upscaling the occupancy map.
7 . The method of claim 6 , wherein reconstructing the volumetric content further comprises obtaining the neural network based on at least one of an indicator representative of a neural network model, a size of block used as input of the neural network or a scale factor by which the occupancy map is upscaled.
8 . The method of claim 7 , wherein the scale factor is determined based on a resolution of the occupancy map and a nominal resolution of the volumetric content.
9 . The method of claim 5 , wherein the information further comprises an indicator representative of a neural network model.
10 . The method of claim 5 , wherein the information further comprises a size of block used as input of a neural network.
11 . The method of claim 5 , wherein a neural network is associated to a second scale factor, and responsive to a determination that the second scale factor does not match a scale factor to which the occupancy map is to be upscaled, upscaling the occupancy map comprises iterating upscaling of the occupancy map using the neural network.
12 . The method of claim 11 , wherein the information further indicates a number of iterations according to which upscaling of the occupancy map is iterated.
13 . The method of claim 5 , wherein the information is signaled at a patch level, or at an image level, or at a sequence level, or in a supplemental enhancement information message, or in a unit header of the bitstream.
14 . The method of claim 5 , wherein the occupancy map comprises at least one value indicating whether at least one sample in a geometry or attribute frame corresponds to at least one associated sample in the volumetric content.
15 . A non-transitory computer readable storage medium having stored thereon instructions for causing one or more processors to perform the method of claim 5 .
16 . An apparatus, comprising one or more processors, wherein the one or more processors are configured to reconstruct a volumetric content from a bitstream, the volumetric content being represented as a set of projections onto one or more atlases, the one or more atlases being video- based encoded, the one or more atlases comprising at least one attribute frame, one geometry frame and one occupancy map, wherein reconstructing the volumetric content comprises:
decoding an information indicating a method among at least two methods for upscaling at least one of the one or more atlases, wherein one of the at least two methods is a neural network-based upscaling,
wherein the at least one of the one or more atlases is an occupancy map,
decoding the occupancy map from the bitstream, and
upscaling the occupancy map.
17 . The apparatus of claim 16 , comprising at least one of (i) an antenna configured to receive a signal, the signal including data representative of at least one part of a volumetric content, (ii) a band limiter configured to limit the signal to a band of frequencies that includes the data representative of the at least one part of the volumetric content, or (iii) a display configured to display the at least one part of the volumetric content.
18 . The apparatus according to claim 17 , comprising a TV, a cell phone, a tablet, a Set Top Box or a Head Mounted Display.
19 . The apparatus of claim 16 , wherein the information indicates use of a neural network for upscaling the occupancy map.
20 . The apparatus of claim 16 , wherein the information further comprises at least one of an indicator representative of a neural network model, a size of block used as input of a neural network or a number of iterations according to which upscaling of the occupancy map is iterated.