HMA

September 26, 2024 · View on GitHub

Updates

✅ 2023-09-09: Release the codes and results of HMA.
✅ 2024-09-26: Release the pre-train models of HMA.

Overview

Benchmark results on SRx4.

Model	Set5	Set14	BSD100	Urban100	Manga109
SwinIR	32.92	29.09	27.92	27.45	32.03
HMA	33.38	29.51	28.13	28.69	33.19

Comparison with the state-of-the-art methods.

Environment

Install Pytorch first. Then,

pip install -r requirements.txt
python setup.py develop

How To Test

Refer to ./options/test for the configuration file of the model to be tested, and prepare the testing data and pretrained model.
The pretrained models are available at Google Drive.
Then run the following codes (taking HMA_SRx2_pretrain.pth as an example):

python hma/test.py -opt options/test/HMA_SRx2.yml

The testing results will be saved in the ./results folder.

How To Train

Refer to ./options/train for the configuration file of the model to train.
Preparation of training data can refer to this page. ImageNet dataset can be downloaded at the official website.
The training command is like

CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 python -m torch.distributed.launch --nproc_per_node=8 --use_env --master_port=4321 hma/train.py -opt options/train/train_HMA_SRx4_from_Imagenet.yml --launcher pytorch

The training logs and weights will be saved in the ./experiments folder.

Results

The inference results on benchmark datasets are available at Google Drive.

Citations

BibTeX

@InProceedings{Chu_2024_CVPR,
author    = {Chu, Shu-Chuan and Dou, Zhi-Chao and Pan, Jeng-Shyang and Weng, Shaowei and Li, Junbao},
title     = {HMANet: Hybrid Multi-Axis Aggregation Network for Image Super-Resolution},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month     = {June},
year      = {2024},
pages     = {6257-6266}
}

Contact

If you have any question, please email douzhichao2021@163.com to discuss with the authors.