Enlighten Anything: When Segment Anything Model Meets Low-Light Image Enhancement

SAM's segment results in low-light images

我们探索了SAM在低光照下的分割结果，如下图所示

Model

weights/ model.py fuse_block.py

Dataset processing

you should run the sam.py to produce the segment image like LOL/LIME/SCIE

Train and Test

python train.py --arc WithoutCalNet --batch_size 10

python test.py

Citations

 @article{Zhao_Zhang_Tang_Gu_Zhu_2023,  
 title={Enlighten-anything:When Segment Anything Model Meets Low-light Image Enhancement}, 
 author={Zhang, Xiaofeng, Qihan Zhao and Tang, Hao and Gu, Chaochen and Zhu, Shanying}, 
 year={2023}, 
 month={Jun}, 
 language={en-US} 
 }