five

renovel/PolarRGB1

收藏
Hugging Face2025-12-09 更新2025-12-20 收录
下载链接:
https://hf-mirror.com/datasets/renovel/PolarRGB1
下载链接
链接失效反馈
官方服务:
资源简介:
--- pretty_name: PolarFree tags: - computer-vision - reflection-removal - polarization - image-processing - cvpr2025 license: cc-by-nc-4.0 datasets: - polarfree language: - en size_categories: - 10K<n<100K task_categories: - image-to-image --- # PolarFree: Polarization-based Reflection-Free Imaging ## Dataset Overview PolarFree is a high-quality dataset designed for polarization-based reflection removal tasks, as introduced in the CVPR 2025 paper "PolarFree: Polarization-based Reflection-Free Imaging". The dataset aims to support tasks such as image reflection removal and image enhancement, particularly suitable for training and evaluating image enhancement models. ## Dataset Structure The dataset is organized as follows: ``` dataset/ ├── train/ │ └── scene_id/ │ ├── input/ │ │ ├── 000_000.png │ │ ├── 000_045.png │ │ └── ... │ └── gt/ │ ├── 000_000.png │ ├── 000_045.png │ └── ... ├── test/ │ └── scene_id/ │ ├── input/ │ │ ├── 000_000.png │ │ ├── 000_045.png │ │ └── ... │ └── gt/ │ ├── 000_000.png │ ├── 000_045.png │ └── ... ``` - **input/scene_id/**: Contains multiple input images captured at different polarization angles. Each group of images shares the same prefix number (e.g., 000, 001), indicating they belong to the same set. - **gt/scene_id/**: Contains the corresponding high-quality ground truth images used as references for the input images. ```json { "inputs": [ "input/scene_id/001_000.png", "input/scene_id/001_045.png", "input/scene_id/001_090.png", "input/scene_id/001_rgb.png" ], "gt": [ "gt/scene_id/000_000.png", "gt/scene_id/000_045.png", "gt/scene_id/000_090.png", "gt/scene_id/000_rgb.png" ] } ``` The training and testing data are all available! If you want raw images, please contact me via mingdeyao@foxmail.com. ## Supported Task - **Image Reflection Removal**: Utilizing polarization information to remove reflections from input images, resulting in clearer images. ## Citation If you use the PolarFree dataset in your research, please cite the following paper: ```bibtex @inproceedings{polarfree2025, title={PolarFree: Polarization-based Reflection-Free Imaging}, author={Yao, Mingde and Wang, Menglu and Tam, King-Man and Li, Lingen and Xue, Tianfan and Gu, Jinwei}, booktitle={CVPR}, year={2025} } ``` For more information, please visit the project page: - GitHub Repository: [https://github.com/mdyao/PolarFree](https://github.com/mdyao/PolarFree) - Paper: [https://arxiv.org/abs/2503.18055](https://arxiv.org/abs/2503.18055) ---
提供机构:
renovel
5,000+
优质数据集
54 个
任务类型
进入经典数据集
二维码
社区交流群

面向社区/商业的数据集话题

二维码
科研交流群

面向高校/科研机构的开源数据集话题

数据驱动未来

携手共赢发展

商业合作