adityassrana / Content-Weighted-Image-Compression

PyTorch implementation of Learning Convolutional Networks for Content-Weighted Image Compression
23 stars 6 forks source link

entropy coding? #1

Closed liujiawei2333 closed 3 years ago

liujiawei2333 commented 3 years ago

Thanks for the code! I want to ask how is the entropy coding stage done? I did not understand the entropy coding part of the original paper. I don’t see this part in your code either.

adityassrana commented 3 years ago

Hi Liu, you can use arithmetic encoding or CABAC implementations available publicly. I did not implement the convolutional entropy encoder mentioned in the paper. I compared my results by calculating the upper limit on the compression rate(bpp) from the importance map.