krasserm/super-resolution
TensorFlow 2 implementation of deep learning models for single image super-resolution using EDSR, WDSR, and SRGAN architectures.
Velocity · 7d
+0.5
★ / day
Trend
→steady
star history
This repository provides implementations of three prominent deep learning architectures for image super-resolution: EDSR (NTIRE 2017 winner), WDSR (NTIRE 2018 winner), and SRGAN (generative adversarial approach). It includes a high-level training API, automatic DIV2K dataset downloading, and fine-tuning capabilities for combining enhancement models with SRGAN for photo-realistic results.