piddnad/DDColor
DDColor is a PyTorch deep learning model that automatically colorizes black-and-white photographs using dual decoder architecture.

Velocity · 7d
+1.2
★ / day
Trend
→steady
star history
DDColor is an official implementation of an ICCV 2023 research paper that provides photo-realistic image colorization. The model uses a dual decoder approach to generate natural, vivid colors for grayscale images including historical photos and anime-style landscapes. It includes pretrained models (DDColor and DDColor-tiny) available on Hugging Face, with support for both inference and training.