Qualcomm Incorporated

Jeffrey Kar Fai Wong

Alireza Shoa Hassani Lashdan

Darren Gnanapragasam

Vladimir Lachine

Stone Yun

Haoping Xu

David Hansen

Sivakumar Chidambaram

Chon-Tam Le Dinh

Dinh Kha Le

Technologies are provided for processing data in neural networks. An example method can include processing, by each layer of a neural network, a row in a first stripe of a data input, the row being processed sequentially in a horizontal direction and according to a layer-by-layer sequence; after processing the row, processing, by each layer, subsequent rows in the first stripe on a row-by-row basis, each subsequent row being processed sequentially in the horizontal direction and according to the layer-by-layer sequence; generating an output stripe based on the processing of the row and subsequent rows; processing, by each layer, a second stripe of the data input, each row in the second stripe being processed in the horizontal direction and according to the layer-by-layer sequence, wherein rows in the second stripe are processed on a row-by-row basis; and generating another output stripe based on the processing of the second stripe.

Processing data in pixel-to-pixel neural networks

Described herein are methods and devices that employ parallel compression of image data using a lossless compressor and a quantization compressor. As described, the lossless compressor generates a variable length compressed bit stream and the quantization compressor generates a fixed length compressed bit stream. The fixed length bit stream is always equal in size to a size requirement of an output memory block. The variable length bit stream may be stored in the output memory block unless it exceeds the memory requirement in size, in which case the fixed length bit stream may be stored to the output memory block.

Image compression method and apparatus for bandwidth saving

Growing community of inventors

Toronto, Canada

Jeffrey Kar Fai Wong