Four-pass perceptual super-resolution with enhanced upscaling
This repository contains a TensorFlow-based implementation of 4PP-EUSR ("Deep learning-based image super-resolution considering quantitative and perceptual quality"), which considers both the quantitative (e.g., PSNR) and perceptual quality (e.g., NIQE) of the upscaled images. Our method won the 2nd place for Region 2 in the PIRM Challenge on Perceptual Super Resolution at ECCV 2018.
Followings are the performance comparison evaluated on the BSD100 dataset.
|Method||PSNR (dB) (↓)||Perceptual Index|
|4PP-EUSR (PIRM Challenge)||26.569||2.683|
Please cite following papers when you use the code, pre-trained models, or results:
- J.-H. Choi, J.-H. Kim, M. Cheon, J.-S. Lee: Deep learning-based image super-resolution considering quantitative and perceptual quality. arXiv:1809.04789 (2018)
- J.-H. Kim, J.-S. Lee: Deep residual network with enhanced upscaling module for super-resolution. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pp. 913-921 (2018)
- Python 3.6+
- TensorFlow 1.8+
Test pre-trained models
Generating upscaled images from the trained models can be done by
Following are the brief instructions.
- Download and copy the trained model available in Downloads section to the
- Place the low-resolution images (PNG only) to the
python test.py --model_name [model file name]. For example, if you downloaded the PIRM Challenge version of our pre-trained model, run
python test.py --model_name 4pp_eusr_pirm.pb.
- The upscaled images will be available on the
python test.py --help for more information.
Currently, the training code is not available. We are working hard on writing the training code for public, so please stay tuned for further updates. :)
Results (Set5, Set14, BSD100, PIRM):