EndoSLAM

该数据集来源于离体猪胃肠道器官，仅包含姿态真值信息。面临的挑战在于只有姿态真值数据。该数据集包含了五个器官，每个器官有五条轨迹，采用五折交叉验证的方式进行规模扩展，其任务是进行姿态估计。This dataset is derived from ex vivo porcine gastrointestinal tract organs and only contains ground-truth pose annotations. The key challenge associated with this dataset is that only ground-truth pose data is provided. It encompasses five organs, with five trajectories available for each organ. A five-fold cross-validation strategy is adopted for dataset scale expansion, and the primary task of this dataset is pose estimation.

搜集汇总

背景与挑战

背景概述

EndoSLAM是一个来源于离体猪胃肠道器官的数据集，仅提供姿态真值信息，专注于姿态估计任务。它包含五个器官，每个器官有五条轨迹，并通过五折交叉验证扩展数据规模，但挑战在于缺乏视觉等辅助数据。