allenai/Molmo2-VideoTrack

Molmo2-VideoTrack是一个视频点跟踪标注数据集，收集了来自16个视频数据集的人类标注数据。该数据集可用于通过点轨迹微调视觉语言模型进行视频对象跟踪。数据集是Molmo2数据集集合的一部分，并用于训练Molmo2系列模型。数据格式包括每个视频剪辑中一个或多个对象的跟踪标注，如唯一标识符、视频文件名、剪辑ID、源数据集名称、文本表达式描述、对象ID列表、点轨迹列表等。

Molmo2-VideoTrack is a dataset of video point tracking annotations collected from human annotators across 16 video datasets. It can be used to fine-tune vision-language models for video object tracking via point trajectories. Molmo2-VideoTrack is part of the Molmo2 dataset collection and was used to train the Molmo2 family of models. Each row contains tracking annotations for one or more objects in a video clip, including fields such as unique identifier, video filename, clip ID, source dataset name, text expression describing the tracked object(s), list of object IDs, list of point trajectories, etc.