SUN-SEG是一个基于知名SUN数据集构建的高质量逐帧标注视频息肉分割(VPS)数据集,包含158,690帧图像。该数据集涵盖多种标注类型,包括物体掩膜(object masks)、边界(boundaries)、涂鸦标记(scribbles)、多边形(polygons)和视觉属性(visual attributes)。此外,完整保留了原始SUN数据集的病理学信息,如病理分类标签(pathological classification labels)、病灶位置细节和形态学特征。
内窥镜维度 | 2D |
模态 | endoscopy |
任务类型 | detection |
解剖结构 | 结肠息肉 |
解剖区域 | 结肠 |
类别数 | 7 |
数据量 | 158,690 |
文件格式 | .jpg, .png |
├──data
├──SUN-SEG
├──TrainDataset
├──Frame # The images from the SUN dataset
├──case1_1
├──image_name_00001.jpg
|...
├──case1_3
|...
├──GT # Object-level segmentation mask
├──case1_1
├──image_name_00001.png
|...
├──case1_3
|...
├──Edge # Weak label with edge
|...
├──Scribble # Weak label with scribble
|...
├──Polygon # Weak label with Polygon
|...
├──Classification # Category classification annotation
├──classification.txt
├──Detection # Bounding box
├──bbox_annotation.json
├──TestEasyDataset
├──Seen
├──Frame
├──case2_3
|...
├──GT
├──case2_3
|...
|...
├──Unseen
├──Frame
├──case3_1
|...
├──GT
├──case3_1
|...
|...
├──TestHardDataset
├──Seen
├──Frame
├──case1_2
|...
├──GT
├──case1_2
|...
|...
├──Unseen
├──Frame
├──case10_1
|...
├──GT
├──case10_1
|...
|...
统计类型 | 间距 (mm) | 尺寸 |
---|---|---|
最小值 | 未提供 |
未提供 |
中位值 | 未提供 |
未提供 |
最大值 | 未提供 |
未提供 |
@article{ji2022vps,
title={Video Polyp Segmentation: A Deep Learning Perspective},
author={Ji, Ge-Peng and Xiao, Guobao and Chou, Yu-Cheng and Fan, Deng-Ping and Zhao, Kai and Chen, Geng and Fu, Huazhu and Van Gool, Luc},
journal={Machine Intelligence Research},
year={2022}
}