FiveCam Dataset Uni-ISP Project Page

The FiveCam Dataset

Synchronized Raw-RGB Captures of Five Smartphone Cameras

The preview of 3 scenes in our new dataset (left) and our capture devices (right). Each scene includes synchronized sRGB-Raw pairs of five smartphone cameras: Apple iPhone 14 Pro Max, Google Pixel 6 Pro, Huawei P40, Samsung Galaxy S20, and Xiaomi Mi 12. The raw images are visualized as XYZ images here, which can be converted back to raw without loss.

Download

We have provided the processed FiveCam dataset that used in training self-camera ISP tasks on 🤗 HuggingFace, with 2217 items for training and 247 items for testing (resolution in 1024*1024).

Preview Self-Camera XYZ-RGB (Test Set) Download Self-Camera XYZ-RGB

Data Format

Following are keys of each data item in self-camera XYZ-RGB datasets:

```
raw_device
```
The name of the device that captures current raw (XYZ) image.
```
rgb_device
```
The name of the device that captures current RGB image. For self-camera dataset, value of `rgb_device`` is the same as the one of `raw_device`.
```
raw_image
```
Resized & cropped XYZ image in PIL.Image format.
```
rgb_image
```
Resized & cropped RGB image in PIL.Image format.
```
raw_exif
```
The EXIF metadata extracted from the original DNG file.
```
rgb_exif
```
The EXIF metadata extracted from the original JPEG file.
```
scene_details
```
The extra metadata of the scene illuminance. Including scene tag, illuminance (lux), illuminance spectral curve, etc

The original Raw-RGB full-resolution version will be available soon.

Citation

😊 Consider citing this paper if you found our dataset useful in your research.

@article{li2024uni,
  title={Uni-ISP: Unifying the Learning of ISPs from Multiple Cameras},
  author={Li, Lingen and Yao, Mingde and Meng, Xingyu and Yu, Muquan and Xue, Tianfan and Gu, Jinwei},
  journal={arXiv preprint arXiv:2406.01003},
  year={2024}
}