Tao Zhang, Xingye Tian, Yu Wu, Shunping Ji, Xuebo Wang...
TL;DR提出一种分离策略,并应用于视频实例分割任务,包括分割、跟踪和细化,使用引用跟踪器和时间细化器构建 Decoupled VIS 框架(DVIS),并在 OVIS 和 VIPSeg 数据集上取得了新的 SOTA 表现。
Abstract
video instance segmentation (VIS) is a critical task with diverse applications, including autonomous driving and video editing. Existing methods often underperform on complex and long videos in real world, primarily due to two factors. Firstly, offline methods are limited by the tightl