ESAD

ESAD数据集是由牛津布鲁克斯大学的视觉人工智能实验室开发的，专门用于微创手术中的外科医生动作检测。该数据集包含四个完整的前列腺切除手术视频，总时长约4小时，共标注了46,325个动作实例。ESAD数据集旨在通过真实测试手术辅助机器人的动作识别能力，提高其在手术中的应用效果。数据集中的动作类别专门针对前列腺切除手术，包括切割、拉扯、夹紧等多种动作。此外，ESAD数据集还被用作2020年MIDL国际会议上的SARAS-ESAD挑战的基础，吸引了众多研究团队参与。

The ESAD dataset was developed by the Visual Artificial Intelligence Laboratory at Oxford Brookes University, specifically for surgeon action detection in minimally invasive surgery. It contains four complete prostatectomy surgery videos, with a total duration of approximately 4 hours and a total of 46,325 annotated action instances. The ESAD dataset aims to test the action recognition capabilities of surgical assistant robots in real clinical scenarios, thereby improving their application effects in surgical operations. The action categories in the dataset are specifically targeted at prostatectomy surgery, including various operative actions such as cutting, retracting, clamping and more. In addition, the ESAD dataset was used as the basis for the SARAS-ESAD Challenge at the 2020 Medical Imaging with Deep Learning (MIDL) International Conference, attracting numerous research teams to participate.