CholecTrack20
收藏资源简介:
该数据集是一个详尽的数据集,它为腹腔镜手术中的多类别多工具追踪进行了精心注释,覆盖了三个视角。该数据集包含20个手术视频,总计超过35,000帧,并标注了65,000个工具实例。为了考虑到追踪中时间连贯性的重要性,该数据集被按照5:1:4的比例分为训练集、验证集和测试集,并主要关注视频而非单帧。规模上,该数据集包含20个视频,超过35,000帧,以及65,000个标注的工具实例,其任务定位于多类别多工具追踪。
This is a comprehensive dataset meticulously annotated for multi-class multi-tool tracking in laparoscopic surgery, covering three camera viewpoints. The dataset includes 20 surgical videos, totaling over 35,000 frames, with 65,000 annotated tool instances. Given the critical role of temporal coherence in tracking tasks, the dataset is divided into training, validation, and test sets at a ratio of 5:1:4, with a primary focus on video-level analysis rather than single-frame processing. In terms of scale, this dataset contains 20 surgical videos, over 35,000 frames, and 65,000 annotated tool instances, targeting the task of multi-class multi-tool tracking.

- 1CholecTrack20: A Dataset for Multi-Class Multiple Tool Tracking in Laparoscopic Surgery斯特拉斯堡大学, 法国国家科学研究中心, 斯特拉斯堡大学医院 · 2023年



