Dive Color Corrector

Restore natural colors in underwater images and videos. Corrects the blue/green color cast caused by water absorbing red light.

Features

• Color correction for underwater images and videos
• Two correction modes: fast hue-shift algorithm or Deep SESR neural network
• GUI application with real-time preview
• CLI for batch processing
• Cross-platform: Windows, Linux

Installation

Download Pre-built Executables

Download from Releases:

Version	Description
Dive Color Corrector	Full version with Deep SESR neural network (~100MB)
Dive Color Corrector Lite	Lightweight version, fast algorithm only (~30MB)

From PyPI

# Lite (GUI only, no neural network)
pip install dive-color-corrector[gui]

# Full (GUI + Deep SESR neural network)
pip install dive-color-corrector[full]

From Source

git clone https://github.com/kcrkor/dive-color-corrector.git
cd dive-color-corrector

uv venv && source .venv/bin/activate

# Lite
uv pip install -e ".[gui]"

# Full (with SESR)
uv pip install -e ".[gui,sesr]"

Usage

GUI Application

python -m dive_color_corrector

Command Line

# Single image
python -m dive_color_corrector image input.jpg output.jpg

# Single image with Deep SESR (requires full install)
python -m dive_color_corrector image input.jpg output.jpg --sesr

# Video
python -m dive_color_corrector video input.mp4 output.mp4

# Batch process directory
python -m dive_color_corrector batch ./raw ./corrected

Examples

Before / After

Sample Video

Algorithm

Two correction methods available:

1. Hue-shift (default): Fast algorithm that shifts hue to restore red channel. Best for most underwater photos.

2. Deep SESR (--sesr flag): Neural network trained on underwater images. Provides super-resolution (2x) and enhanced color correction. Slower but can produce better results for challenging images.

Development

# Install dev dependencies
uv pip install -e ".[dev,gui,sesr]"

# Run tests
pytest tests/

# Run linting
ruff check src tests
mypy src

License

MIT License - see LICENSE

Acknowledgments

• Algorithm inspired by nikolajbech/underwater-image-color-correction
• Thanks to bornfree and all contributors