TY - GEN
T1 - End-To-End Learned Image Compression with Fixed Point Weight Quantization
AU - Sun, Heming
AU - Cheng, Zhengxue
AU - Takeuchi, Masaru
AU - Katto, Jiro
N1 - Funding Information:
This work was supported in part by JST, PRESTO Grant Number JP-MJPR19M5, Japan, and in part by Hoso Bunka Foundation.
Publisher Copyright:
© 2020 IEEE.
PY - 2020/10
Y1 - 2020/10
N2 - Learned image compression (LIC) has reached the traditional hand-crafted methods such as JPEG2000 and BPG in terms of the coding gain. However, the large model size of the network prohibits the usage of LIC on resource-limited embedded systems. This paper presents a LIC with 8-bit fixed-point weights. First, we quantize the weights in groups and propose a non-linear memory-free codebook. Second, we explore the optimal grouping and quantization scheme. Finally, we develop a novel weight clipping fine tuning scheme. Experimental results illustrate that the coding loss caused by the quantization is small, while around 75% model size can be reduced compared with the 32-bit floating-point anchor. As far as we know, this is the first work to explore and evaluate the LIC fully with fixed-point weights, and our proposed quantized LIC is able to outperform BPG in terms of MS-SSIM.
AB - Learned image compression (LIC) has reached the traditional hand-crafted methods such as JPEG2000 and BPG in terms of the coding gain. However, the large model size of the network prohibits the usage of LIC on resource-limited embedded systems. This paper presents a LIC with 8-bit fixed-point weights. First, we quantize the weights in groups and propose a non-linear memory-free codebook. Second, we explore the optimal grouping and quantization scheme. Finally, we develop a novel weight clipping fine tuning scheme. Experimental results illustrate that the coding loss caused by the quantization is small, while around 75% model size can be reduced compared with the 32-bit floating-point anchor. As far as we know, this is the first work to explore and evaluate the LIC fully with fixed-point weights, and our proposed quantized LIC is able to outperform BPG in terms of MS-SSIM.
KW - Image compression
KW - fine-tuning
KW - fixed-point
KW - neural networks
KW - quantization
UR - http://www.scopus.com/inward/record.url?scp=85098644580&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85098644580&partnerID=8YFLogxK
U2 - 10.1109/ICIP40778.2020.9190805
DO - 10.1109/ICIP40778.2020.9190805
M3 - Conference contribution
AN - SCOPUS:85098644580
T3 - Proceedings - International Conference on Image Processing, ICIP
SP - 3359
EP - 3363
BT - 2020 IEEE International Conference on Image Processing, ICIP 2020 - Proceedings
PB - IEEE Computer Society
T2 - 2020 IEEE International Conference on Image Processing, ICIP 2020
Y2 - 25 September 2020 through 28 September 2020
ER -