Orbuculum – Predicting When Users Intend to Leave Large Public Displays
summary
A system predicts when a user is about to leave a large public display by detecting behavioral features from video and Microsoft Kinect skeleton data; 240 videos are coded to find cues like foot movement, body rotation and relaxation, and a deployed Quiz game (N=249) uses a decision tree to trigger an intervention (raise an arm) and achieves 78% accuracy. Beginning to walk to the side leaves about 3.5 seconds on average for detection; an initial lab set uses 44 labeled users to train the model.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:32Contributions
- 2:00Methodology
- 3:55Interaction Flow
- 4:23Accuracy
- 4:43Next Steps
- 5:24Conclusion
related talks
10mProactive Sensing for Improving Hand Pose Estimation
08mSo Predictable! Continuous 3D Hand Trajectory Prediction in Virtual Reality
22mSeeing, Sensing and Recognizing Laban Movement Qualities
11mDetection and Differentiation of Obstacles in Repeated Adaptive Human-Computer Interactions from ...
19mGroup Touch: Distinguishing Tabletop Users in Group Settings via Statistical Modeling of Touch Pairs
18mTo Explore or To Exploit: Decoding Human Decision Making with R and Python
12mThe Potential of Cognitive Circles to Measure Mental Load
39mMind-Reading with Adaptive and Intelligent User Interfaces