🎨 Prism - Image Colorization
Transform grayscale images into vibrant color photos using deep learning!
How to use:
- Upload a grayscale or color image
- The model will convert it to grayscale and predict the colors
- View the colorized result
Model: PRISM (PyTorch U-Net based colorization model)
Input Image
Output Image
Try Example (Grayscale)
Examples
About
This model was trained on image colorization using a U-Net architecture with:
- Grayscale input (L channel)
- Color prediction in LAB color space (313 discrete color classes)
- Weighted cross-entropy loss for balanced color prediction
Authors
- Nishant Kalaichelvan - @nkminion
- Varun Agnihotri - @PythonicVarun - hello@pythonicvarun.me
License
See the LICENSE file for more details.