Continuous Optical Zooming Dataset

Our dataset is availble at COZ - Google Drive.

Local Mix Implicit Network for Arbitrary-Scale Image Super-Resolution (LMI)

Official PyTorch implementation of LMI network.

Installation

Our code is based on Ubuntu 20.04, pytorch 1.11.0, CUDA 11.3 (NVIDIA RTX 3090 24GB, NVIDIA A40 48GB) and python 3.8.

Data Preparation

Download our dataset and unzip it in the current directory.

Train & Test

EDSR-Baseline-LMI

Train: python train_real.py --config configs/train-real/lmi-edsr-baseline.yaml

Test: python test_real.py --config configs/test/test-RealAbrSR.yaml --model save/LMI-edsr-baseline/epoch-best.pth

RDN-LMI

Train: python train_real.py --config configs/train-real/lmi-rdn.yaml

Test: python test_real.py --config configs/test/test-RealAbrSR.yaml --model save/LMI-rdn/epoch-best.pth

We use NVIDIA RTX 3090 24GB for training, and NVIDIA A40 48GB for testing.

Pretrained Checkpoints

EDSR-Baseline-LMI

RDN-LMI

Acknowledgements

This code is built on LIIF and LTE. We thank the authors for sharing their codes.

Citation

If you find our work useful in your research, please consider citing our paper:

@InProceedings{Fu_2024_CVPR,
    author    = {Fu, Huiyuan and Peng, Fei and Li, Xianwei and Li, Yejun and Wang, Xin and Ma, Huadong},
    title     = {Continuous Optical Zooming: A Benchmark for Arbitrary-Scale Image Super-Resolution in Real World},
    booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
    month     = {June},
    year      = {2024},
    pages     = {3035-3044}
}