Collaborative knowledge distillation for incomplete multi-view action prediction

作者：

Highlights：

• Multi-view methods assume complete observation and fuse both views

• Early event detection should be made on incomplete observation

• Missing view or partial view blocked by object can occur any time

• Graph Attention + Knowledge Distillation to mitigate the corrupted data

• Ensemble learning to predict action from unseen data corruption

摘要

•Multi-view methods assume complete observation and fuse both views•Early event detection should be made on incomplete observation•Missing view or partial view blocked by object can occur any time•Graph Attention + Knowledge Distillation to mitigate the corrupted data•Ensemble learning to predict action from unseen data corruption

论文关键词：Multi-view,Action prediction,Knowledge distillation,Graph attention

论文评审过程：Received 8 November 2020, Revised 7 January 2021, Accepted 11 January 2021, Available online 21 January 2021, Version of Record 5 February 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104111