Last Friday, my advisor asked why the entropy coding in my encoding method takes twice as long as another method. My intuitive understanding was that his method used an optimized entropy coding C++ library, rather than directly using the entropy coding provided by CompressAI.

So yesterday I spent the entire day adapting the entropy coding part of my method to his optimized entropy coding. I found that the entropy coding speed is exactly the same! ＞﹏＜

Today I spent another day investigating the reason, analyzing from multiple angles such as whether the bpp is similar, to understand why the entropy coding in my method is slower. The result!! The result is that the input data volume for my entropy coding is more than 3 times his!

What a shame, my two wasted days.