Apparatus and method for transceiving feature map extracted using MPEG-VCM

A neural processing unit (NPU) for encoding video or feature map, the NPU comprising:

• a first circuitry provided for at least one processing element (PE) to perform an inference using an artificial neural network, the at least one PE is configured to receive and encode a transmitted video or feature map,
• wherein the at least one PE is configured to output a bitstream,
• wherein the bitstream includes:
• data of a base layer;
• or
• the data of the base layer and data of at least one enhancement layer,
• wherein the data of the base layer included in the bitstream includes a first feature map, and the data of the at least one enhancement layer included in the bitstream includes a second feature map,
• wherein the bitstream is transmitted in unit of one frame,
• wherein the NPU receives feedback information on a number of enhancement layers in the frame from a decoding apparatus, and
• wherein the number of enhancement layers is determined based on at least one of a transmission channel and a machine analysis task of the decoding apparatus.

An apparatus for decoding video and/or feature map, the apparatus comprising:

• a first circuitry provided for at least one processing element (PE) to perform an inference using an artificial neural network, the at least one PE is configured to receive and decode data included in a bitstream,
• wherein the data included in the bitstream comprises:
• data of a base layer, or
• the data of the base layer and data of at least one enhancement layer,
• wherein the data of the base layer included in the bitstream includes a first feature map, and the data of the at least one enhancement layer included in the bitstream includes a second feature map,
• wherein the bitstream is received in unit of one frame,
• wherein a number of enhancement layers to be included in the frame is determined based on at least one of a transmission channel and a machine analysis task, and
• wherein feedback information on the number of enhancement layers is transmitted to an encoding apparatus.

An apparatus for encoding video and/or feature map the apparatus comprising:

• a first circuitry provided for at least one processing element (PE) to perform an inference using an artificial neural network, the at least one PE is configured to receive and encode a transmitted video or feature map,
• wherein the at least one PE is configured to output a bitstream,
• wherein the bitstream includes:
• data of a base layer;
• or
• the data of the base layer and data of at least one enhancement layer,
• wherein the data of the base layer included in the bitstream includes a first feature map, and the data of the at least one enhancement layer included in the bitstream includes a second feature map,
• wherein the bitstream is transmitted in unit of one frame,
• wherein an NPU receives feedback information on a number of enhancement layers in the frame from a decoding apparatus, and
• wherein the number of enhancement layers is determined based on at least one of a transmission channel and a machine analysis task of the decoding apparatus.