Efficient Long-Short Temporal Attention network for unsupervised Video Object Segmentation

Ping Li, Yu Zhang, Li Yuan, Huaxin Xiao, Binbin Lin, Xianghua Xu

Abstract: Highlights•Efficient Long-Short Temporal Attention network (LSTA) is developed for unsupervised video object segmentation.•The Long Temporal Memory (LTM) module captures the long-term global pixel relations.•The Short Temporal Attention (STA) module reveals the short-term local pixel relations.•It achieves nearly linear time complexity for the two light modules.•Empirical studies verify the promising speed-accuracy tradeoff of the proposed method.