该数据集名为YouTube-VIS 2021，相较于2019版本，它在训练视频数量上有所增加，具体包括2985个训练视频、421个验证视频以及453个测试视频，同时保持了40个对象类别。此外，该数据集的标注质量相较于YouTube-VIS 2019有了显著提升。在规模上，它涵盖了40个对象类别，以及上述提到的视频数量。该数据集的任务是对视频实例进行分割。

arXiv1030

MIVE Dataset

多实例视频编辑

视频实例分割

The MIVE (Multi-Instance Video Editing) Dataset is a new dataset for multi-instance video editing tasks. It contains 200 videos with varying numbers and sizes of instances accompanied with instance-le

IEEE100

Quantitative results for different metrics under varying levels of distance between instances are presented in terms of mIoU on the Youtube-VIS 2019 dataset.

视频实例分割

图像分割性能评估

Quantitative results for different metrics under varying levels of distance between instances are presented in terms of mIoU on the Youtube-VIS 2019 dataset.

Figshare2024-10-07 更新40

Quantitative results for different metrics under varying levels of occlusion in terms of mIoU on the Youtube-VIS 2019 dataset.

资源简介：

Quantitative results for different metrics under varying levels of occlusion in terms of mIoU on the Youtube-VIS 2019 dataset.

资源简介：

相关数据集