--- title: dddd canonical_url: "https://www.modelscope.cn/datasets/donglanyixue/dddd" md_url: "https://www.modelscope.cn/datasets/donglanyixue/dddd.md" repository: donglanyixue/dddd chinese_name: div last_updated: 2025-04-23 license: "Apache License 2.0" storage_size: "5.0 MB" downloads: 3 stars: 0 --- # dddd > dddd - donglanyixue 在 ModelScope 开源的数据集。Abstract: Transformers have exhibited promising performance in computer vision tasks including image super-resolution (SR). However, popular transformer-based SR methods often employ window self-attention with… donglanyixue/dddd 是 ModelScope 魔搭社区上的数据集，存储大小 5.0 MB，采用 Apache License 2.0 许可。 - **Repository**: donglanyixue/dddd - **License**: Apache License 2.0 - **Storage size**: 5.0 MB - **Downloads**: 3 - **Stars**: 0 - **Last updated**: 2025-04-23 Source: https://www.modelscope.cn/datasets/donglanyixue/dddd --- > **Abstract:** *Transformers have exhibited promising performance in computer vision tasks including image super-resolution (SR). However, popular transformer-based SR methods often employ window self-attention with quadratic computational complexity to window sizes, resulting in fixed small windows with limited receptive fields. In this paper, we present a general strategy to convert transformer-based SR networks to hierarchical transformers (HiT-SR), boosting SR performance with multi-scale features while maintaining an efficient design. Specifically, we first replace the commonly used fixed small windows with expanding hierarchical windows to aggregate features at different scales and establish long-range dependencies. Considering the intensive computation required for large windows, we further design a spatial-channel correlation method with linear complexity to window sizes, efficiently gathering spatial and channel information from hierarchical windows. Extensive experiments verify the effectiveness and efficiency of our HiT-SR, and our improved versions of SwinIR-Light, SwinIR-NG, and SRFormer-Light yield state-of-the-art SR results with fewer parameters, FLOPs, and faster speeds (~7x).*

## 📑 Contents - [🔥News](#-News) - [🛠️Setup](#%EF%B8%8F-Setup) - [💿Datasets](#-Datasets) - [🚀Models](#-Models) - [🏋Training](#-Training) - [🧪Testing](#-Testing) - [📊Results](#-Results) - [📎Citation](#-Citation) - [🏅Acknowledgements](#-Acknowledgements) --- ## 🔥 News - 2025-03: 🚀The DF2K version of HiT-SRF ([HiT-SRF-DF2K](#-Models)) is released! - 2024-09: 🤗HiT-SR is available at [🤗Hugging Face](https://huggingface.co/XiangZ/hit-sr). Thank [Niels](https://github.com/NielsRogge)! - 2024-08: 🧑‍💻HiT-SRF is available at [neosr](https://github.com/muslll/neosr). Thank [muslll](https://github.com/muslll)! - 2024-07: 🎉HiT-SR is accepted by ECCV 2024! This repo is released. ## 🛠️ Setup - Python 3.8 - PyTorch 1.8.0 + Torchvision 0.9.0 - NVIDIA GPU + [CUDA](https://developer.nvidia.com/cuda-downloads) ```bash git clone https://github.com/XiangZ-0/HiT-SR.git conda create -n HiTSR python=3.8 conda activate HiTSR pip install -r requirements.txt python setup.py develop ``` ## 💿 Datasets Training and testing sets can be downloaded as follows: | Training Set | Testing Set | Visual Results | | :-----------------------------------------------------------: | :----------------------------------------------------------: | :----------------------------------------------------------: | | [DIV2K](https://data.vision.ee.ethz.ch/cvl/DIV2K/) (800 training images, 100 validation images) [organized training dataset DIV2K: [One Drive](https://1drv.ms/u/c/de821e161e64ce08/Eb1dyRMuCJBGjmtUUJd1j2EBbDhcSyHBYqUeqKjhuPb49Q?e=3RMxbs)] | Set5 + Set14 + BSD100 + Urban100 + Manga109 [complete testing dataset: [One Drive](https://1drv.ms/u/c/de821e161e64ce08/EUN4kTCUdBtNuvJnb2Jy3BkByBMErLIqpiQI4NG6HcAXWQ?e=3k5dGK)] | [One Drive](https://1drv.ms/f/c/de821e161e64ce08/EuE6xW-sN-hFgkIa6J-Y8gkB9b4vDQZQ01r1ZP1lmzM0vQ?e=hV5OOc) | A larger training dataset DF2K ([DIV2K](https://data.vision.ee.ethz.ch/cvl/DIV2K/) + [Flickr2K](https://cv.snu.ac.kr/research/EDSR/Flickr2K.tar)) can also be used for better performance [DF2K: [One Drive](https://1drv.ms/u/c/de821e161e64ce08/EfSn064NEU5AjF1BjfdqxVgBnfj28TK2Bfceg6oD0T8Imw?e=re2BPH)] Please download training and testing datasets and put them into the corresponding folders of `datasets/`. See [datasets](datasets/README.md) for the detail of the directory structure. ## 🚀 Models | Method | #Param. (K) | FLOPs (G) | Dataset | PSNR (dB) | SSIM | Model Zoo | Visual Results | | :-------- | :----: | :-------: | :------: | :-------: | :----: | :----------------------------------------------------------: | :----------------------------------------------------------: | | HiT-SIR | 792 | 53.8 | Urban100 (x4) | 26.71 | 0.8045 | [One Drive](https://1drv.ms/f/c/de821e161e64ce08/EhLkXZsiGV9HgjwjQIvNV3oBTKmSaTZfZ0-jIMMJtONN3w?e=PSos9v) | [One Drive](https://1drv.ms/u/c/de821e161e64ce08/Eeya10xIX-dJsGVa3WJt2hkBnTeG3CTJFuP9tLdwBHBndg?e=fb1aM4) | | HiT-SNG | 1032 | 57.7 | Urban100 (x4) | 26.75 | 0.8053 | [One Drive](https://1drv.ms/f/c/de821e161e64ce08/ElBD_V3wgy9KrotqdSoWyQoB2BhOUzcPxYkFQyQQp68jYA?e=Kz4LRw) | [One Drive](https://1drv.ms/u/c/de821e161e64ce08/Ee6a-XKo1qFKrTvgOiFlb4sBfNfyLBwHnMVj-vqfxO5YRA?e=pURhUB) | | HiT-SRF | 866 | 58.0 | Urban100 (x4) | 26.80 | 0.8069 | [One Drive](https://1drv.ms/f/c/de821e161e64ce08/ErtsTu3cbxdFnVPFJAcofY4BkwfGq5c0pGewFIBNTkujrg?e=wLd1n1) | [One Drive](https://1drv.ms/u/c/de821e161e64ce08/ET9b9T7PdDdGr8T7EFdX8OkBzK3vBe1drGD-LAcyGYgr-g?e=aYGwOP) | | HiT-SRF-DF2K | 866 | 58.0 | Urban100 (x4) | 27.00 | 0.8119 | [One Drive](https://1drv.ms/f/c/de821e161e64ce08/El8rHwr9naRMmAptyq4k02oB2wqKGodgaIDQ38heMQvATA?e=8pa6Gg) | [One Drive](https://1drv.ms/u/c/de821e161e64ce08/EaGJK29f5QBMi8msg3Vl5xkB_CUbzRlHGI1cxRj3jIi2qQ?e=9p3GUn) | The output size is set to 1280x720 to compute FLOPs. The performance of HiT-SRF-DF2K is (PSNR/SSIM): | Method | Scale | Set5 | Set14 | B100 | Urban100 | Manga109 | |---|---|---|---|---|---|---| | HiT-SRF-DF2K | x2 | 38.30/0.9615 | 34.06/0.9217 | 32.41/0.9027 | 33.30/0.9387 | 39.67/0.9793 | | HiT-SRF-DF2K | x3 | 34.79/0.9301 | 30.68/0.8486 | 29.33/0.8113 | 29.16/0.8717 | 34.71/0.9510 | | HiT-SRF-DF2K | x4 | 32.63/0.8993 | 28.96/0.7899 | 27.78/0.7442 | 27.00/0.8119 | 31.55/0.9203 | ## 🏋 Training - Download training ([DIV2K](https://1drv.ms/u/c/de821e161e64ce08/Eb1dyRMuCJBGjmtUUJd1j2EBbDhcSyHBYqUeqKjhuPb49Q?e=3RMxbs) or [DF2K](https://1drv.ms/u/c/de821e161e64ce08/EfSn064NEU5AjF1BjfdqxVgBnfj28TK2Bfceg6oD0T8Imw?e=re2BPH), already processed) and [testing](https://1drv.ms/u/c/de821e161e64ce08/EUN4kTCUdBtNuvJnb2Jy3BkByBMErLIqpiQI4NG6HcAXWQ?e=3k5dGK) (Set5, Set14, BSD100, Urban100, Manga109, already processed) datasets, place them in `datasets/`. - Run the following scripts. The training configuration is in `options/Train/`. ```shell # HiT-SIR, input=64x64, 4 GPUs python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SIR_x2.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SIR_x3.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SIR_x4.yml --launcher pytorch # HiT-SNG, input=64x64, 4 GPUs python -m torch.distributed.launch --nproc_per_node=4 --master_port=4321 basicsr/train.py -opt options/Train/train_HiT_SNG_x2.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=4321 basicsr/train.py -opt options/Train/train_HiT_SNG_x3.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=4321 basicsr/train.py -opt options/Train/train_HiT_SNG_x4.yml --launcher pytorch # HiT-SRF, input=64x64, 4 GPUs python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SRF_x2.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SRF_x3.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SRF_x4.yml --launcher pytorch # HiT-SRF-DF2K, input=64x64, 4 GPUs python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SRF_x2_DF2K.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SRF_x3_DF2K.yml --launcher pytorch python -m torch.distributed.launch --nproc_per_node=4 --master_port=1234 basicsr/train.py -opt options/Train/train_HiT_SRF_x4_DF2K.yml --launcher pytorch ``` - The training experiments will be stored in `experiments/`. ## 🧪 Testing ### Test with ground-truth images - Download the [pre-trained models](https://1drv.ms/f/c/de821e161e64ce08/EqakXUlsculBpo79VKpEXY4B_6OQL-fGyilrzpHaNObG1A?e=YNrqHb) and place them in `experiments/pretrained_models/`. We provide pre-trained models for efficient image SR: HiT-SIR, HiT-SNG, and HiT-SRF (x2, x3, x4). - Download [testing datasets](https://1drv.ms/u/c/de821e161e64ce08/EUN4kTCUdBtNuvJnb2Jy3BkByBMErLIqpiQI4NG6HcAXWQ?e=3k5dGK) (Set5, Set14, BSD100, Urban100, Manga109), place them in `datasets/`. - Run the following scripts. The testing configuration is in `options/Test/` (e.g., [test_HiT_SIR_x2.yml](options/Test/test_HiT_SIR_x2.yml)). Note 1: You can set `use_chop: True` (default: False) in YML to chop the image for testing. ```shell # No self-ensemble # HiT-SIR, reproduces results in Table 2 of the main paper python basicsr/test.py -opt options/Test/test_HiT_SIR_x2.yml python basicsr/test.py -opt options/Test/test_HiT_SIR_x3.yml python basicsr/test.py -opt options/Test/test_HiT_SIR_x4.yml # HiT-SNG, reproduces results in Table 2 of the main paper python basicsr/test.py -opt options/Test/test_HiT_SNG_x2.yml python basicsr/test.py -opt options/Test/test_HiT_SNG_x3.yml python basicsr/test.py -opt options/Test/test_HiT_SNG_x4.yml # HiT-SRF, reproduces results in Table 2 of the main paper python basicsr/test.py -opt options/Test/test_HiT_SRF_x2.yml python basicsr/test.py -opt options/Test/test_HiT_SRF_x3.yml python basicsr/test.py -opt options/Test/test_HiT_SRF_x4.yml # HiT-SRF-DF2K, reproduces results in the above Models section python basicsr/test.py -opt options/Test/test_HiT_SRF_x2_DF2K.yml python basicsr/test.py -opt options/Test/test_HiT_SRF_x3_DF2K.yml python basicsr/test.py -opt options/Test/test_HiT_SRF_x4_DF2K.yml ``` - The output is stored in `results/`. All visual results of our pre-trained models can be accessed via [one drive](https://1drv.ms/f/c/de821e161e64ce08/EuE6xW-sN-hFgkIa6J-Y8gkB9b4vDQZQ01r1ZP1lmzM0vQ?e=aIRfCQ). ### Test without ground-truth images - Download the [pre-trained models](https://1drv.ms/f/c/de821e161e64ce08/EqakXUlsculBpo79VKpEXY4B_6OQL-fGyilrzpHaNObG1A?e=YNrqHb) and place them in `experiments/pretrained_models/`. We provide pre-trained models for efficient image SR: HiT-SIR, HiT-SNG, and HiT-SRF (x2, x3, x4). - Put your dataset (single LR images) in `datasets/single`. Some example images are in this folder. - Run the following scripts. The testing configuration is in `options/test/` (e.g., [test_single_x2.yml](options/Test/test_single_x2.yml)). Note 1: The default model is HiT-SRF. You can use other models like HiT-SIR by modifying the YML. Note 2: You can set `use_chop: True` (default: False) in YML to chop the image for testing. ```shell # Test on your dataset without ground-truth images python basicsr/test.py -opt options/Test/test_single_x2.yml python basicsr/test.py -opt options/Test/test_single_x3.yml python basicsr/test.py -opt options/Test/test_single_x4.yml ``` - The output is stored in `results/`. ## 📊 Results We apply our HiT-SR approach to improve [SwinIR-Light](https://github.com/JingyunLiang/SwinIR), [SwinIR-NG](https://github.com/rami0205/NGramSwin) and [SRFormer-Light](https://github.com/HVision-NKU/SRFormer), corresponding to our HiT-SIR, HiT-SNG, and HiT-SRF. Compared with the original structure, our improved models achieve better SR performance while reducing computational burdens. - Performance improvements of HiT-SR (SIR, SNG, and SRF indicate SwinIR-Light, SwinIR-NG, and SRFormer-Light, respectively).

- Efficiency improvements of HiT-SR (SIR, SNG, and SRF indicate SwinIR-Light, SwinIR-NG, and SRFormer-Light, respectively). The complexity metrics are calculated under x2 upscaling on an A100 GPU, with the output size set to 1280x720.

- Overall improvements of HiT-SR

- Convergence improvements of HiT-SR

More detailed results can be found in the paper. All visual results of can be downloaded [here](https://1drv.ms/f/c/de821e161e64ce08/EuE6xW-sN-hFgkIa6J-Y8gkB9b4vDQZQ01r1ZP1lmzM0vQ?e=aIRfCQ).
More results (click to expan) - Quantitative comparison

- [Local attribution map (LAM)](https://x-lowlevel-vision.github.io/lam.html) comparison (more marked pixels indicate better information aggragation ability)

- Qualitative comparison on challenging scenes

## 📎 Citation If you find the code helpful in your research or work, please consider citing the following paper. ``` @inproceedings{zhang2024hitsr, title={HiT-SR: Hierarchical Transformer for Efficient Image Super-Resolution}, author={Zhang, Xiang and Zhang, Yulun and Yu, Fisher}, booktitle={ECCV}, year={2024} } ``` ## 🏅 Acknowledgements This project is built on [DAT](https://github.com/zhengchen1999/DAT), [SwinIR](https://github.com/JingyunLiang/SwinIR), [NGramSwin](https://github.com/rami0205/NGramSwin), [SRFormer](https://github.com/HVision-NKU/SRFormer), and [BasicSR](https://github.com/XPixelGroup/BasicSR). Special thanks to their excellent works!