Conventional video segmentations usually use low-level features to extract object.

• 传统的视频分割通常利用低级特征提取对象。