Group members: Loretta Cheung, Irina Bocharnikova, Yuri Allasia, Mahmoud Alaa Salem

Webcam-Based Learner Engagement Estimation

This demo estimates observable engagement proxies from face visibility, eye openness, gaze-related signals, and facial behaviour. It does not claim to read true emotions.

Model loading...

Live detection

Face visible —

Eyes open —

Gaze toward screen proxy —

Expression tension proxy —

Focus score —

Learner state

Waiting

Start the webcam. The state will update after several frames.

Session

Session time: 0:00

Adaptive feedback

No feedback yet. The app waits for sustained patterns before showing suggestions.

Project note

This is a demo page of a project of CCNU summer school, under the topic 'Learner State Estimation'. Please notice that this demo page describe these outputs as visual engagement indicators, not reliable emotion detection. The important research question is whether such indicators can support gentle adaptive learning feedback.