Patent · US Active

Method and system for multimodal analysis based emotion recognition

US11386712B2 · kind B2 · utility

0Cited by
1References
20Claims
0Family size

Assignee

Inventors

Key dates

Filing dateFeb 20, 2020
Grant dateJul 12, 2022
Priority date
Expiry dateJan 22, 2041

Classification

  • Technology area (CPC G)Physics
  • CPC primaryG06F40/30
  • WIPO fieldComputer technology
  • WIPO sectorElectrical engineering

Abstract

The present invention discloses method and system for multimodal analysis based emotion recognition. The method comprising segmenting video data of a user into a plurality of video segments. A plurality of visual features, voice features and text features from the plurality of video segments is extracted. Autocorrelation values among each of the plurality of visual features, the voice features, and the text features is determined. Each of the plurality of visual features, the voice features and the text features is aligned based on video segment identifier and the autocorrelation values to obtain a plurality of aligned multimodal features. One of two classes of emotions is determined for each of the plurality of aligned multimodal features. The determined emotion for each of the plurality of aligned multimodal features is compared with historic multimodal features from a database, and emotion of the user is determined at real time based on comparison.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.