SCARED数据集源自2019年MICCAI会议的内窥镜单目深度估计挑战赛(Endoscopic Monocular Depth Estimation Challenge),由Intuitive Surgical公司发布。该数据集已成为内窥镜深度估计领域使用最广泛的基准和训练数据集。
内窥镜维度 | 2D |
模态 | endoscopy |
任务类型 | other |
解剖结构 | 腹部 |
解剖区域 | 腹腔 |
数据量 | 280G |
文件格式 | tiff, png, mp4 |
SCARED/
├── dataset_1/
│ ├── keyframe_1/
│ │ ├── data/
│ │ │ ├── scene_points.tar.gz
│ │ │ ├── rgb.mp4
│ │ │ └── frame_data.tar.gz
│ │ ├── right_depth_map.tiff
│ │ ├── left_depth_map.tiff
│ │ ├── point_cloud.obj
│ │ ├── Left_Image.png
│ │ ├── Right_Image.png
│ │ └── endoscope_calibration.yaml
│ ├── keyframe_2/
│ │ └── [similar structure as keyframe_1]
│ ├── keyframe_3/
│ │ └── [similar structure as keyframe_1]
│ ├── keyframe_4/
│ │ └── [similar structure as keyframe_1]
│ └── keyframe_5/
│ └── [similar structure as keyframe_1]
统计类型 | 间距 (mm) | 尺寸 |
---|---|---|
最小值 | - |
1280x1024 |
中位值 | - |
1280x1024 |
最大值 | - |
1280x1024 |
@article{allan2021stereo,
title={Stereo correspondence and reconstruction of endoscopic data challenge},
author={Allan, Max and Mcleod, Jonathan and Wang, Congcong and Rosenthal, Jean Claude and Hu, Zhenglei and Gard, Niklas and Eisert, Peter and Fu, Ke Xue and Zeffiro, Trevor and Xia, Wenyao and others},
journal={arXiv preprint arXiv:2101.01133},
year={2021}
}