Bimodal Emotion Recognition for the Patients with Depression

Xuesong Wang, Shenghui Zhao*, Yingxue Wang

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

With the rapid development of the society, over three hundred million people from worldwide suffer from depression, which has become one of the most serious health problems in the world. On the other hand, emotion recognition research works barely focuses on the depression patients although their emotion differs from that of the normal people obviously. Firstly, a speech/text emotion dataset was built up under the dialogue scenes with both the depressed patients and the normal people, and the speech/text fragments were classified into five common emotions: sadness, anger, happy, fear and neutral. Secondly, a bimodal emotion recognition algorithm is proposed, which uses a multimodal Transformer model as the feature fusion module. The experimental results show that it achieves an accuracy of 69.2% for the normal people and 59.4% for the depression patients.

Original languageEnglish
Title of host publication2021 6th International Conference on Signal and Image Processing, ICSIP 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages40-43
Number of pages4
ISBN (Electronic)9780738133737
DOIs
Publication statusPublished - 2021
Event6th International Conference on Signal and Image Processing, ICSIP 2021 - Nanjing, China
Duration: 22 Oct 202124 Oct 2021

Publication series

Name2021 6th International Conference on Signal and Image Processing, ICSIP 2021

Conference

Conference6th International Conference on Signal and Image Processing, ICSIP 2021
Country/TerritoryChina
CityNanjing
Period22/10/2124/10/21

Keywords

  • Depression
  • Emotion recognition
  • Feature fusion
  • Multimodal transformer

Fingerprint

Dive into the research topics of 'Bimodal Emotion Recognition for the Patients with Depression'. Together they form a unique fingerprint.

Cite this