Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/reshalfahsi/zero-reference-low-light-image-enhancement

Zero-Reference Low-Light Image Enhanchement
https://github.com/reshalfahsi/zero-reference-low-light-image-enhancement

image-processing low-light-image-enhancement pytorch pytorch-lightning zero-dce

Last synced: about 6 hours ago
JSON representation

Zero-Reference Low-Light Image Enhanchement

Host: GitHub
URL: https://github.com/reshalfahsi/zero-reference-low-light-image-enhancement
Owner: reshalfahsi
Created: 2023-07-22T00:03:37.000Z (over 1 year ago)
Default Branch: master
Last Pushed: 2023-07-31T15:40:28.000Z (over 1 year ago)
Last Synced: 2023-07-31T16:07:11.698Z (over 1 year ago)
Topics: image-processing, low-light-image-enhancement, pytorch, pytorch-lightning, zero-dce
Language: Jupyter Notebook
Homepage:
Size: 10.3 MB
Stars: 0
Watchers: 1
Forks: 0
Open Issues: 0
Metadata Files:
- Readme: README.md

Awesome Lists containing this project

README

# Zero-Reference Low-Light Image Enhancement

Low-light image enhancement aims to raise the quality of pictures taken in dim lighting, resulting in a brighter, clearer, and more visually appealing image without adding too much noise or distortion. One of the state-of-the-art methods for this computer vision task is Zero-DCE. This method uses just a low-light image without any image reference to learn how to produce an image with higher brightness. There are four loss functions crafted specifically for this zero-reference low-light image enhancement method, i.e., color constancy loss, exposure loss, illumination smoothness loss, and spatial consistency loss.

## Experiment

Open the following [link](https://github.com/reshalfahsi/zero-reference-low-light-image-enhancement/blob/master/Zero_Reference_Low_Light_Image_Enhancement.ipynb) and hit the run all in the colab notebook to examine the overall processes.

## Result

## Quantitative Result

The quantitative performance of the model is exhibited in the table below.

Metrics | Test Dataset |
------------ | ------------- |
Color Constancy Loss | 0.065 |
Exposure Loss | 0.391 |
Illumination Smoothness Loss | 0.094 |
Spatial Consistency Loss | 0.042 |
Total Loss | 0.592 |
PSNR | 13.646 |
SSIM | 0.663 |
MAE | 0.170 |

## Evaluation Metrics Curve

colorconstancy_loss_curve
Color constancy loss curve on the train set and the validation set.

exposure_loss_curve
Exposure loss curve on the train set and the validation set.

illumination_smoothness_loss_curve
Illumination smoothness loss curve on the train set and the validation set.

spatialconsistency_loss_curve
Spatial consistency loss curve on the train set and the validation set.

totalloss_curve
Total loss curve on the train set and the validation set.

psnr_curve
PSNR curve on the train set and the validation set.

ssim_curve
SSIM curve on the train set and the validation set.

mae_curve
MAE curve on the train set and the validation set.

## Qualitative Result

Here are some samples of the qualitative results of the model.

enhancement_qualitative_00
enhancement_qualitative_01
enhancement_qualitative_02
The qualitative results of the image enhancement method (comparing the original, the ground-truth, the PIL autocontrast, and the prediction).

## Credit

- [(Tutorial) Zero-DCE for low-light image enhancement](https://keras.io/examples/vision/zero_dce/)
- [(Paper) Zero-Reference Deep Curve Estimation for Low-Light Image Enhancement](https://arxiv.org/pdf/2001.06826.pdf)
- [PyTorch Lightning](https://lightning.ai/docs/pytorch/latest/)