EndoVis 2017 Challenge

该数据集名为EndoVis 2017挑战数据集，提供了手术过程中立体相机拍摄的视频资料，遵循原始比赛指南，并排除了手术器械中的“其他标签”类别。在评估时，主要以平均交并比（mIoU）作为主要评价指标，并遵循原始比赛的规定。该数据集共包含8组数据，每组225帧，任务是对手术器械进行无监督分割。

This dataset, named the EndoVis 2017 Challenge Dataset, provides video footage captured by stereo cameras during surgical procedures. It follows the original competition guidelines and excludes the "other" label category for surgical instruments. For evaluation, mean Intersection over Union (mIoU) is adopted as the primary evaluation metric, in accordance with the regulations of the original competition. This dataset consists of 8 subsets, each containing 225 frames, and the task is unsupervised segmentation of surgical instruments.