wuyuheng/how2compress repository overview

This is the official implementation of "How2Compress: Efficient Edge Video Analytics via Adaptive Granular Video Compression" (MM2025).

docker run -it --gpus all -e "NVIDIA_DRIVER_CAPABILITIES=all" -e "NVIDIA_VISIBLE_DEVICES=all" --network host -v /<user_dir>/dataset:/dataset --shm-size=188g --name how2compress wuyuheng/how2compress

if you encounter any issue of reproducibility, please email wyhforwork66(at)gmail(dot)com or you can raise issue on https://github.com/wyhallenwu/how2compress⁠.