WildLive Dataset

该数据集名为WildLive，包含了215,800个边界框和动物分割遮罩，以及291个针对斑马、长颈鹿和大象的追踪轨迹。这些数据是从22段由无人机获取的4K视频序列中提取出来的，总帧数达到19,139帧。此外，该数据集还包括经验证的动物边界框、分割遮罩（SAM2）、追踪轨迹ID以及使用DJI Mavic 3无人机收集的手动校正稀疏LK像素轨迹。规模上，数据集跨越19,000多帧，拥有超过200,000个标注的动物实例。该数据集的任务是野生动物的动物检测与追踪。

This dataset, named WildLive, contains 215,800 bounding boxes and animal segmentation masks, as well as 291 tracking trajectories for zebras, giraffes and elephants. All data are extracted from 22 4K video sequences captured by drones, totaling 19,139 frames. Additionally, the dataset includes validated animal bounding boxes, segmentation masks (SAM2), tracking trajectory IDs, and manually corrected sparse LK pixel trajectories collected using the DJI Mavic 3 drone. In terms of scale, the dataset spans over 19,000 frames and contains more than 200,000 annotated animal instances. The core task of this dataset is wildlife animal detection and tracking.