GazeChat: Enhancing Virtual Conferences with Gaze Awareness and Interactive 3D Photos
summary
System uses a single webcam with WebGazer (smoothed by a 1€ filter) plus optional Tobii input to estimate gaze, synthesizes 20 eye-angle frames with a pre-trained model, infers depth to make a 3D photo, swaps in gaze images, and ships gaze events over WebRTC/WebSocket. Calibration expects ~80% accuracy and face >25% frame. A remote study (N=16) compares audio with static photos, GazeChat ThirdPerson vs EyeContact, and video, finding higher engagement, more perceived eye contact, and lower bandwidth than full video.
AI-written summary · about
chapters (6)
- 0:00Introduction to gaze chat
- 1:03Previous research context
- 2:19System overview
- 3:18Implementation details
- 6:36User study and evaluation
- 7:37Limitations and future work
related talks
17mGaze Augmentation in Egocentric Video Improves Awareness of Intention
15mCan Eye Help You?: Effects of Visualizing Eye Fixations on Remote Collaboration Scenarios for ...
20mWeb based 3D Camera Simulator for Perceptual Computing
19mAggreGaze: Collective Estimation of Audience Attention on Public Displays
32mDesigning Beyond the Screen
22mEye tracking the user experience – should you?
12mHolographic video calling
42mEyes are the windows to our souls