ShadowRemovalRefine

Shadow Removal Refinement via Material-Consistent Shadow Edges.

Shilin Hu, Hieu Le, ShahRukh Athar, Sagnik Das, Dimitris Samaras. WACV 2025. [link]

Proposed Benchmark Testset

400 pairs of test images and annotated edges. GoogleDrive

Getting Started

conda env create -f environment.yml

Fine-tuned SAM

You can directly use the checkpoint. Or train your own by

python train_finetunedSAM.py

Run ShadowRemovalRefine

First, change line83 in refine.py to your data path.

Then, run test_time_adaptation.sh

Our results are on GoogleDrive

Baseline Models

Check ShadowFormer and SP+M-Net for baseline models.