Main objective
Develop an automatic emotion recognition system, based on visual, vocal and textual signals.
Context
- Under CIFRE contract between the LIMOS laboratory and Jeolis Solutions digital service company
- Main application cases:
- Adapt software content for a motivational purpose: personalised physical activity coaching (e.g., obesity), as part of patient education.
- Estimate the emotional state from a video support to assist health professionals in remote mental health monitoring.
Main Challenges
- Multimodal fusion of heterogeneous and high dimensional data
- Manage conflicting information across and within modalities
- Capture the ambiguity around emotion
Papers accepted
Completed and Ongoing Tasks
Related to my thesis
- Learn the main concepts of emotional psychology
- Define the application cases of the thesis with respect to the company’s projects
- Literature review of multimodal databases for emotion recognition
- Literature review of multimodal emotion recognition models
- Experimentation phase: development of an innovative emotion recognition model
- Bi-monthly meeting with the laboratory and the company teams
Other PhD activities
- Collaborate with three master students on emotion recognition
- Communication manager of Miners LinkedIn page (Data Mining research group of LIMOS)
- Manage the contents of the Miners website
- Participate in the organization of scientific days for doctoral students (June 2022)
Useful Links
Miners team website Jeolis Solutions website (R&D)
Miners team (Data Mining Research Group)