Video-based emotion recognition in the wild using deep transfer learning and score fusion
作者：
Highlights：
• We present transfer learning strategies for robust emotion recognition in the wild.
• We compare and contrast a set of visual descriptors and video modeling methods.
• We propose a small but effective set of summarizing functionals for video modeling.
• We compare feature and score level fusion alternatives.
• We report state-of-the-art results on EmotiW, Chalearn LAP FI, and CK+ corpora.
摘要
•We present transfer learning strategies for robust emotion recognition in the wild.•We compare and contrast a set of visual descriptors and video modeling methods.•We propose a small but effective set of summarizing functionals for video modeling.•We compare feature and score level fusion alternatives.•We report state-of-the-art results on EmotiW, Chalearn LAP FI, and CK+ corpora.
论文关键词：EmotiW,Emotion recognition in the wild,Multimodal fusion,Convolutional neural networks,Kernel extreme learning machine,Partial least squares
论文评审过程：Received 16 April 2016, Revised 21 January 2017, Accepted 26 January 2017, Available online 4 February 2017, Version of Record 18 September 2017.
论文官网地址：https://doi.org/10.1016/j.imavis.2017.01.012
