FudanCVL/MOVE
收藏Hugging Face2025-09-07 更新2025-09-13 收录
下载链接：
https://hf-mirror.com/datasets/FudanCVL/MOVE
下载链接
链接失效反馈官方服务：
资源简介：
MOVE是一个大规模的数据集，专为运动引导的少样本视频对象分割任务设计。它包含了基于相同运动模式的少量标注样本，旨在对视频中的动态对象进行分割。该数据集弥补了现有数据集和方法在理解视频中的丰富时间动态方面的不足，有助于推动对运动理解有要求的应用场景的研究。
MOVE is a large-scale dataset specifically designed for motion-guided few-shot video object segmentation. It contains a small number of annotated examples based on the same motion patterns, aiming to segment dynamic objects in videos. The dataset addresses the limitations of existing datasets and methods in capturing the rich temporal dynamics in videos, contributing to research in scenarios that require motion understanding.
提供机构：
FudanCVL
