Low-Light-Enhancement

Low light image enhancement using MIRNet in PyTorch.

Check out my blog for thorough understanding.

Reults:

This model achieves 22.97 PSNR on LoL dataset's evaluation set.
Input image:point_right: Restored/Predicted Image :point_right: Ground Truth image

Details:

1. This model uses LoL dataset.
2. All the images were resized to 64x64 during training.
3. It uses 3 Residual Recurrent Groups(RRGs) where every RRG contains 2 Multi-scale residual blocks(MSRBs).
4. You can see more results in Results directory.