What it does This repo is a meticulously organized reading list and link index for deep-learning image and video colorization. It separates fully automatic methods from user-guided ones—whether the user supplies scribbles, reference images, color palettes, or text prompts—and tracks which papers ship code or demos.

The interesting bit The taxonomy itself is the contribution. The maintainer slices the field into unusual cross-sections like “Human-AI Collaborated Colorization System” and even distinguishes colorizing natural photos from filling flat colors in manga line art. It is a literature review that stays current through pull requests.

Key highlights

• Covers automatic, user-guided, and video colorization with sub-categories for scribble, palette, text, and reference-image control
• Links to code or project pages for most entries, including well-known tools like DeOldify and Palette.fm
• Spans 2015 to 2024 (and one 2026 entry), showing the field’s evolution from GANs to diffusion transformers
• Includes niche domains like sketch and manga colorization, not just natural photographs
• Accepts community contributions via GitHub issues and pull requests

Caveats

• The README is a long list of tables with no synthesis, analysis, or quality ranking—you still have to read the papers
• Some entries lack code links, and the “Awesome” badge promises more curation rigor than is visibly enforced

Verdict Researchers and practitioners who need to survey the colorization landscape fast should bookmark this. If you want a pre-trained model or a tutorial, look elsewhere.