Play and rewind: Context-aware video temporal action proposals
作者：
Highlights：
• We propose a novel context-aware temporal action proposal generation network for TAP task. It makes full use of the contextual information during both initial proposal generation and re-ranking phases.
• For generating initial proposals, we design a Bi-directional Parallel LSTMs to extract the visual features of a video unit by considering its contextual information.
• For re-ranking the proposals, we design an action-attention based re-ranking network which considers both surrounding proposal and initial actionness scores.
• Extensive experiments on the challenging datasets demonstrate the effectiveness of our proposed framework.
摘要
•We propose a novel context-aware temporal action proposal generation network for TAP task. It makes full use of the contextual information during both initial proposal generation and re-ranking phases.•For generating initial proposals, we design a Bi-directional Parallel LSTMs to extract the visual features of a video unit by considering its contextual information.•For re-ranking the proposals, we design an action-attention based re-ranking network which considers both surrounding proposal and initial actionness scores.•Extensive experiments on the challenging datasets demonstrate the effectiveness of our proposed framework.
论文关键词：Temporal action proposal generation and detection,Deep learning,Untrimmed video analysis
论文评审过程：Received 28 May 2019, Revised 3 May 2020, Accepted 28 May 2020, Available online 6 June 2020, Version of Record 11 June 2020.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107477
