Learning for Video Compression with Hierarchical Quality and Recurrent Enhancement(CVPR 2020) - Video Compression Paper Reading

Hierarchical Learned Video Compression(HLVC) method: 3 hierarchical quality layers and a recurrent enhancement network.
First layer: compress by an image conpression method with the highest quality.
Second layer: Bi-Directional Deep Compression(BDDC) network, relatively high quality.
Third layer: Single Motion Deep Compression(SMDC) network, lowest quality.
Code: https://github.com/RenYang-home/HLVC