RainDrop / DeRaindrop: Lens-Adherent Raindrop Removal Dataset (Unofficial Mirror)

Unofficial redistribution of RainDrop (also known as DeRaindrop), the lens-adherent raindrop removal dataset from Qian et al. (CVPR 2018), packaged for direct use with ClearView's dataset pipeline. This is ClearView's first raindrop removal dataset, a physically distinct degradation from rain streaks or general haze.

Disclaimer

This repository is not an official release of RainDrop/DeRaindrop.

RainDrop was created by Rui Qian, Robby T. Tan, Wenhan Yang, Jiajun Su, and Jiaying Liu. This repository does not claim ownership of any images or metadata, and attributes the dataset to its original creators. (We describe them as "creators" rather than "copyright holders" deliberately, copyright in academic datasets can rest with an author's university, a funding body, or otherwise, under arrangements we have no way to verify from the outside.)

On licensing, please read before relying on this mirror. We looked for a formal license covering this dataset (project page, code repository, paper supplementary material) and found none, the original distribution is a plain Google Drive link with no accompanying terms of use. This is the same situation as this collection's Rain13K and DDN-Data mirrors: no formal license found anywhere, but no stated restriction either (contrast this with a dataset that explicitly says "research purposes only", which we treat differently and do not mirror). In the absence of an explicit license, we have chosen to preserve this dataset for research reproducibility while providing full attribution and making the mirror removable on request, this is a decision we're making under that uncertainty, not a claim that we've established a legal right to redistribute it. Concretely:

• Full attribution and citation to the original creators (below).
• No claim of any rights beyond hosting a copy.
• If the copyright holder, or an authorized representative, requests removal, we will comply promptly.

If you need certainty about redistribution rights (e.g. for commercial use), contact the dataset's creators or their institution directly rather than relying on this mirror.

---

Dataset Overview

RainDrop is a real-world raindrop removal benchmark, physically distinct from rain-streak deraining: instead of streaks falling through a scene, it captures droplets adhered to a glass pane or camera lens. The authors captured pairs using two panes of glass side by side, one clean, one sprayed with water, so the background scene stays spatially aligned between each pair.

• Training: 861 pairs.
• Testing (A): 58 pairs, .png.
• Testing (B): 249 pairs, .jpg (a superset benchmark; Test-A is commonly used for well-aligned qualitative comparison, Test-B for the fuller quantitative benchmark).

Note: some downstream papers cite Test-B at 239 pairs; this mirror's actual file count is 249, verified directly against the official download rather than the commonly-repeated figure.

---

Changes from the Official Release

None beyond repackaging. No images added, removed, or modified. No relabeling. The train/{data,gt}, test_a/{data,gt}, and test_b/{data,gt} directory layout matches the official archive's structure exactly.

---

Dataset Structure

raindrop/
├── README.md
├── banner.jpg
├── train/
│   ├── data/    # {id}_rain.png  (861 images)
│   └── gt/      # {id}_clean.png (861 images)
├── test_a/
│   ├── data/    # {id}_rain.png  (58 images)
│   └── gt/      # {id}_clean.png (58 images)
└── test_b/
    ├── data/    # {id}_rain.jpg  (249 images)
    └── gt/      # {id}_clean.jpg (249 images)

Rainy and clean images share a common numeric ID but different filename suffixes ({id}_rain.ext / {id}_clean.ext), not identical stems, this is why ClearView ships a dedicated RainDropDataset parser rather than relying on the generic ImagePairDataset (which expects exact-stem matches).

Split	Pairs	Format
train	861	.png
test_a	58	.png
test_b	249	.jpg
Total	1,168

---

Usage with ClearView

This dataset is designed to be used directly with ClearView, an open-source PyTorch framework for image restoration.

from huggingface_hub import snapshot_download
from clearview.data import RainDropDataset, get_train_transforms, get_val_transforms

data_dir = snapshot_download(repo_id="dronefreak/RainDrop", repo_type="dataset")

train_ds = RainDropDataset(root_dir=data_dir, split="train", transform=get_train_transforms(crop_size=(256, 256)))
test_a_ds = RainDropDataset(root_dir=data_dir, split="test_a", transform=get_val_transforms())
test_b_ds = RainDropDataset(root_dir=data_dir, split="test_b", transform=get_val_transforms())

rainy, clean = train_ds[0]

Or directly via the training CLI:

clearview-train \
  --data-dir <path-to-downloaded-snapshot> \
  --dataset-type pair \
  --train-rainy train/data --train-clean train/gt \
  --val-rainy test_a/data --val-clean test_a/gt \
  --model unet --batch-size 16 --crop-size 256 --epochs 100 \
  --loss l1 --mixed-precision \
  --output-dir ./runs/raindrop_unet_baseline

Note: --dataset-type pair maps to the generic ImagePairDataset, which expects identical filename stems and will not match this dataset's {id}_rain/{id}_clean suffix convention out of the box. Use the Python API above (RainDropDataset) for correct pairing until a dedicated CLI --dataset-type raindrop is wired in.

---

Dataset Sources

Original Paper

Attentive Generative Adversarial Network for Raindrop Removal from a Single Image

Rui Qian, Robby T. Tan, Wenhan Yang, Jiajun Su, Jiaying Liu

IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2018.

• arXiv: https://arxiv.org/abs/1711.10098
• CVPR Open Access (PDF): https://openaccess.thecvf.com/content_cvpr_2018/papers/Qian_Attentive_Generative_Adversarial_CVPR_2018_paper.pdf

Official Resources

• Code + Dataset: https://github.com/rui1996/DeRaindrop

---

Attribution

All credit for the dataset belongs entirely to the original authors: Rui Qian, Robby T. Tan, Wenhan Yang, Jiajun Su, and Jiaying Liu.

If you use this dataset in your research, please cite the original publication below.

---

License

No formal license was located for this dataset, see the Disclaimer above for our reasoning for mirroring it anyway under that uncertainty (full attribution, reproducibility purpose, removable on request). This repository makes no claim to any rights over the dataset content beyond hosting a copy.

---

Citation

If you use this dataset, please cite:

@InProceedings{Qian_2018_CVPR,
  author = {Qian, Rui and Tan, Robby T. and Yang, Wenhan and Su, Jiajun and Liu, Jiaying},
  title = {Attentive Generative Adversarial Network for Raindrop Removal From a Single Image},
  booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  month = {June},
  year = {2018}
}

---

Acknowledgements

We sincerely thank Rui Qian, Robby T. Tan, Wenhan Yang, Jiajun Su, and Jiaying Liu for creating and publicly releasing this foundational raindrop-removal benchmark.