Retinexnet: enhance low-light images from the command line

Understand the project

Practical overview

Use Retinexnet when you need a scriptable, local workflow for enhancing low-light images or want to experiment with training on paired datasets.

Developers building local image-processing workflows that need a low-light enhancement step.
Researchers or practitioners experimenting with model training on paired low-light image data.

Process images from a directory and save enhanced results.
Choose whether test outputs include decomposition results as well as enhanced images.
Train the model using LOL and synthetic paired datasets.

Before you choose

The repository is public and lists an MIT license.

Requires Python, TensorFlow 1.5 or later, NumPy and PIL.
Training requires paired datasets arranged in the repository's documented folders.
The README says enhancement performance is sensitive to training parameters.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Run the repository's test phase on images in a directory and save the enhanced results.

Set the test phase and provide the input directory with --test_dir.
Choose an output directory with --save_dir.
Set --decom=0 to save only enhanced results, or use the documented option to include decomposition results.
Run the command below with the selected arguments.
Documented command
```
python main.py
```

The script saves image results to the chosen output directory.

The README's demo command saves results under ./test_results/; it does not specify a separate setup or installation procedure.
Enhancement performance depends on parameters, and changed defaults may produce poor results.

Evidence and transparency

Source record details

Source availability: Active
Sources checked: 17 October 2026
Original website title: weichen582/RetinexNet · Official homepage title · retained in its source language