Consumer Stereo Video Segmentation Challenge (CSVSC) dataset
收藏arXiv2016-09-07 更新2024-08-06 收录
下载链接：
http://arxiv.org/abs/1609.00836v2
下载链接
链接失效反馈官方服务：
资源简介：
CSVSC数据集是由马克斯·普朗克信息学研究所创建的，旨在推动消费者立体视频分割的研究。该数据集包含30个从YouTube挑选的异质视频序列，涵盖了多种对象数量、演员类型、运动类型及视觉复杂度，以及不同的相机参数和噪声。数据集的创建过程涉及对视频的精心选择和人类标注，以确保数据的质量和多样性。CSVSC数据集主要用于视频分割技术的评估和改进，特别是在处理动态和静态场景中的对象分割问题。
The CSVSC Dataset was developed by the Max Planck Institute for Informatics, aiming to advance research on consumer stereoscopic video segmentation. This dataset includes 30 heterogeneous video sequences selected from YouTube, covering diverse object quantities, actor types, motion patterns, visual complexities, as well as varying camera parameters and noise levels. The dataset creation process involves careful video screening and human annotation to guarantee data quality and diversity. The CSVSC Dataset is primarily used for the evaluation and optimization of video segmentation technologies, particularly for object segmentation tasks in both dynamic and static scenes.
提供机构：
马克斯·普朗克信息学研究所创建时间：
2016-09-03
