https://github.com/dbpprt/duts-hq
HQ version of the DUTS dataset for saliency detection
https://github.com/dbpprt/duts-hq
Last synced: 3 months ago
JSON representation
HQ version of the DUTS dataset for saliency detection
- Host: GitHub
- URL: https://github.com/dbpprt/duts-hq
- Owner: dbpprt
- Created: 2022-02-22T19:01:12.000Z (over 4 years ago)
- Default Branch: main
- Last Pushed: 2022-03-25T14:02:36.000Z (about 4 years ago)
- Last Synced: 2025-11-15T11:03:41.466Z (7 months ago)
- Size: 8.58 MB
- Stars: 5
- Watchers: 1
- Forks: 0
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
DUTS-HQ
HQ version of the [DUTS](http://saliencydetection.net/duts/) dataset for saliency detection, upscaled using [SwinIR](https://github.com/JingyunLiang/SwinIR) and refined by utilising [CascadePSP](https://github.com/hkchengrex/CascadePSP).
---
## Features
- All images are upscaled by 4x using SwinIR and the ```003_realSR_BSRGAN_DFOWMFC_s64w8_SwinIR-L_x4_GAN.pth``` pretrained model
- The images are square padded to 1600x1600 with black borders and all masks upscaled using bicubic interpolation
- The masks are refined using [CascadePSP](https://github.com/hkchengrex/CascadePSP)
- This dataset **only** includes the DUTS train set and not the validation set
- The dataset contains the upscaled masks in the ```masks``` subdirectory and the processed masks in the ```hq-masks``` subdirectory
- All images are stored as PNG, 3 channel for the images and 1 channel for the masks
- All original filenames are retained
## Samples
Image
Mask
HQ Mask
### Original images
Image
Mask
## Download links
- [GH](https://github.com/dennisbappert/duts-hq/releases/tag/dataset)
## Warning
The refined masks aren't always perfect, but they are pretty good. There are a couple of quite noisy samples in the dataset, hence I included the original upscaled and padded masks in the dataset.
## Citation
```bibtex
@inproceedings{wang2017,
title={Learning to Detect Salient Objects with Image-level Supervision},
author={Wang, Lijun and Lu, Huchuan and Wang, Yifan and Feng, Mengyang
and Wang, Dong, and Yin, Baocai and Ruan, Xiang},
booktitle={CVPR},
year={2017}
}
```
```bibtex
@inproceedings{liang2021swinir,
title={SwinIR: Image Restoration Using Swin Transformer},
author={Liang, Jingyun and Cao, Jiezhang and Sun, Guolei and Zhang, Kai and Van Gool, Luc and Timofte, Radu},
booktitle={IEEE International Conference on Computer Vision Workshops},
year={2021}
}
```
```bibtex
@inproceedings{cheng2020cascadepsp,
title={{CascadePSP}: Toward Class-Agnostic and Very High-Resolution Segmentation via Global and Local Refinement},
author={Cheng, Ho Kei and Chung, Jihoon and Tai, Yu-Wing and Tang, Chi-Keung},
booktitle={CVPR},
year={2020}
}
```
## License and Acknowledgement
This dataset is based on the initial work of ```wang2017``` augmented using SwinIR by ```liang2021swinir``` and CascadePSP by ```cheng2020cascadepsp``` and follows their original licenses.