What AI says about BSRGAN

BSRGAN is an open-source blind image super-resolution framework that uses a GAN to reconstruct high-resolution images from real-world degraded inputs by learning a practical degradation model. The repository provides training scripts, pretrained models, and utilities for end-to-end SR under realistic degradation conditions. It is developed by researchers Kai Zhang, Jingyun Liang, Luc Van Gool, and Radu Timofte from ETH Zurich, with notes on research updates and paper milestones.