Non-volume preserving-based fusion to group-level emotion recognition on crowd videos

作者：

Highlights：

• Estimating group-level emotions on crowd videos using fused facial features.

• An effective facial expression network to extract facial expression features.

• Deep feature-level fusion mechanism combining spatial and temporal features.

• Collecting a large video database for group-level emotion on crowd videos.

摘要

•Estimating group-level emotions on crowd videos using fused facial features.•An effective facial expression network to extract facial expression features.•Deep feature-level fusion mechanism combining spatial and temporal features.•Collecting a large video database for group-level emotion on crowd videos.

论文关键词：Group-level emotion recognition,Facial features,Feature extraction,Feature fusion,Crowd videos

论文评审过程：Received 7 July 2020, Revised 7 February 2022, Accepted 8 March 2022, Available online 18 March 2022, Version of Record 28 March 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108646