US 12,132,919 B2
Neural image compression with controllable spatial bit allocation
Yang Yang, San Diego, CA (US); Hoang Cong Minh Le, La Jolla, CA (US); Yinhao Zhu, La Jolla, CA (US); Reza Pourreza, San Diego, CA (US); Amir Said, San Diego, CA (US); Yizhe Zhang, San Diego, CA (US); and Taco Sebastiaan Cohen, Amsterdam (NL)
Assigned to QUALCOMM Incorporated, San Diego, CA (US)
Filed by QUALCOMM Incorporated, San Diego, CA (US)
Filed on Nov. 15, 2022, as Appl. No. 17/987,844.
Claims priority of provisional application 63/280,097, filed on Nov. 16, 2021.
Prior Publication US 2023/0156207 A1, May 18, 2023
Int. Cl. H04N 19/124 (2014.01); H04N 19/119 (2014.01); H04N 19/147 (2014.01); H04N 19/17 (2014.01); H04N 19/436 (2014.01)
CPC H04N 19/436 (2014.11) [H04N 19/119 (2014.11); H04N 19/124 (2014.11); H04N 19/147 (2014.11); H04N 19/17 (2014.11)] 20 Claims
OG exemplary drawing
 
1. A processor-implemented method, comprising:
receiving, at an encoder of an artificial neural network, an image and a spatial segmentation map corresponding to the image, the spatial segmentation map indicating one or more regions of interest;
learning, at an auxiliary neural network, a quantization bin size based on the spatial segmentation map;
generating a learned quantization bin size in accordance with the learning; and
compressing, via the encoder, the image according to a controllable spatial bit allocation, the controllable spatial bit allocation is based on the learned quantization bin size.