Method, apparatus and system for encoding and decoding tensors
The method addresses CNN computational complexity by encoding and decoding tensors with unexecuted batch normalization, ensuring efficient compression and accuracy in collaborative intelligence systems.
JP2025532752AInactive Publication Date: 2025-10-03CANON KK
View PDF 0 Cites 0 Cited by
Patent Information
- Application Number
- JP2025507409
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2022-10-13
- Filing Date
- 2023-07-28
- Publication Date
- 2025-10-03
- Estimated Expiration
- Not applicable · inactive patent
Smart Images
Figure 2025532752000001_ABST
Abstract
A system and method for encoding a tensor associated with image data into a bitstream includes obtaining a first tensor for the image data, the first tensor being derived using a portion of a neural network, the neural network including at least a plurality of layers of a first type, each layer of the first type having at least a convolution module and a batch normalization module, the first tensor corresponding to a tensor for which the convolution module of one of the plurality of layers of the first type has been executed but the batch normalization module of one of the plurality of layers of the first type has not been executed. The method further includes performing a predetermined process on the first tensor to derive a second tensor and encoding the second tensor into a bitstream. The number of dimensions of the data structure of the first tensor is greater than the number of dimensions of the data structure of the second tensor.
Need to check novelty before this filing date? Find Prior Art