Raindrop - xyfJASON/image-datasets GitHub Wiki

Links

Official website | Papers with Code | Google drive

Brief introduction

Copied from paperswithcode.

Raindrop is a set of image pairs, where each pair contains exactly the same background scene, yet one is degraded by raindrops and the other one is free from raindrops. To obtain this, the images are captured through two pieces of exactly the same glass: one sprayed with water, and the other is left clean. The dataset consists of 1,119 pairs of images, with various background scenes and raindrops. They were captured with a Sony A6000 and a Canon EOS 60.

Statistics

Numbers: 1,110

Splits: 861 / 58 / 249 (train / test_a / test_b)

Resolution: ~720x480

Note: test_a is a subset of test_b where the alignment of image pairs is good.

Usage

File structure

Please organize the downloaded dataset in the following file structure:

root
├── train
│   ├── data
│   │   ├── 0_rain.png
│   │   ├── ...
│   │   └── 860_rain.png
│   ├── gt
│   │   ├── 0_clean.png
│   │   ├── ...
│   │   └── 860_clean.png
│   └── preview.html
├── test_a
│   ├── data
│   │   ├── 0_rain.png
│   │   ├── ...
│   │   └── 57_rain.png
│   └── gt
│       ├── 0_clean.png
│       ├── ...
│       └── 57_clean.png
└── test_b
    ├── data
    │   ├── 0_rain.jpg
    │   ├── ...
    │   └── 248_rain.jpg
    └── gt
        ├── 0_clean.jpg
        ├── ...
        └── 248_clean.jpg

API Reference

Raindrop(root: str, split: str = 'train', transforms: Optional[Callable] = None)

• root: Root directory of dataset.
• split: One of {'train', 'test'}.
• transforms: A function/transform that takes in degraded and ground-truth images and returns the transformed version.

Example

from image_datasets import Raindrop

root = '~/data/Raindrop/'  # path to downloaded dataset
train_set = Raindrop(root=root, split='train')
testa_set = Raindrop(root=root, split='test_a')
testb_set = Raindrop(root=root, split='test_b')
print(len(train_set))  # 861
print(len(testa_set))  # 58
print(len(testb_set))  # 249
print(train_set[0])    # (<PIL.Image.Image image mode=RGB size=720x480 at 0x7FAAEF6C8BD0>, <PIL.Image.Image image mode=RGB size=720x480 at 0x7FAAF10DFB90>)