Extremely low-bitrate Image Compression Semantically Disentangled by LMMs from a Human Perception Perspective
April 11, 2025 ยท View on GitHub
Result
Dependency
Preparation
Installation
A suitable conda environment named SEDIC can be created and activated with:
conda env create -n CL-LRPE python=3.8.1
conda activate CL-LRPE
Install environment
pip install -r requirements.txt
Weights
Download weights and put them into the weight folder:
DiffBIR (general_full_v1.ckpt): link
Cheng2020-Tuned (cheng_small.pth.tar): link
GroundingDINO (GroundingDINO-T): link
SAM (sam_vit_h_4b8939.pth): link
test
You can test this on your dataset by running the following command:
python SEDIC.py