Endoscapes

Endoscapes数据集由法国斯特拉斯堡大学ICube实验室创建，专注于腹腔镜胆囊切除术（LC）视频的自动安全关键视图评估。该数据集包含201个LC视频，总计11090帧带有安全关键视图（CVS）标注和1933帧带有工具及解剖结构边界框标注。数据集的创建过程涉及多轮专家标注和审核，确保标注的一致性和准确性。Endoscapes数据集的应用领域主要集中在手术场景分割、对象检测和CVS评估，旨在通过自动化技术提高手术安全性。The Endoscapes dataset was developed by the ICube Laboratory of the University of Strasbourg, France, focusing on automated safety-critical view assessment for laparoscopic cholecystectomy (LC) videos. This dataset contains 201 LC videos, with a total of 11,090 frames annotated with safety-critical views (CVS) and 1,933 frames annotated with bounding boxes for surgical instruments and anatomical structures. The dataset construction process involves multiple rounds of expert annotation and review to ensure annotation consistency and accuracy. The main application scenarios of the Endoscapes dataset cover surgical scene segmentation, object detection and CVS assessment, aiming to improve surgical safety via automated technologies.