Please go to ["./Sun_glint_detection/dataset/sun_glint/README.md"] to prepare the dataset.
Please go to ["./Sun_glint_detection/utils/mypath.py] to modify the path.
Please go to ["./Sun_glint_contamination_restoration/data/README.md"] to prepare the dataset.
Please go to google drive to download the weights.
Please prepare an environment with python=3.7, and then use the command "pip install -r requirements.txt" for the dependencies.
- Run the train script on sun glint dataset.
cd Sun_glint_detection
python train.py --dataset sun_glint --use_balanced_weights --model fassnet
- If you have multiple GPUs, run:
python train.py --dataset sun_glint --use_balanced_weights --model fassnet --sync-bn True --gpu_ids 0,1
- To evaluate the performance of FASSNet, run:
python test.py --dataset sun_glint --model fassnet --sync-bn True --gpu_ids 0,1
cd Sun_glint_contamination_restoration
cd tool
python sunglint_restoration.py --path ../data/glint --path_mask ../data/glint_mask --outroot ../result/glint