Human Computer Intelligent Interaction Thomas S. Huang Department of Electrical and Computer Engineering Coordinated Science Laboratory Beckman Institute for Advanced Science and Technology Center for Advanced Study University of Illinois at Urbana-Champaign
Four broad research areas 1. Biometrics 2. Video event detection and recognition 3. Human computer interaction 4. Web-scale multimedia processing and applications
Human Computer Interaction Multimodal Processing Display; Engine Avatar Sensors Physical Human Environment Memory Robotics Multimodal Interface Computer Human Studies: Perception, Cognition, Behavior
Examples: Human to Computer Hand tracking and applications to manipulating virtual objects, etc. Shrug detection Soft Biometrics: Recognition of gender, age group, emotion Audio-visual speech recognition
Example: Computer to Human Audio-Visual emotive avatar
Mona Lisa was a white female, 43 year old, 83% happy
Images Data: Video Range 3D Face Model Geometry Texture Motion Analysis (tracking) Synthesis (animation) Lip reading Text- and Speech- driven talking face Emotion recognition
Facial Movement Action units ~10 Muscles Control points ~100 Smoothness Mesh vertices ~1000
Comparison of bimodal ASR and acoustic ASR recognition performances
Future: Biometrics Recognizing gender, age group, emotion/ affect from non-frontal faces and body/ movement
Future: Robust Real-Time Algorithms Sensors: High-speed cameras; 3D cameras; multiple cameras Parallel computing architectures Clever algorithms
Future: Applications Finding applications to fit algorithm performance: Human in the loop Multimodality Forgiving
Demonstrations By Zhen Li Gender and age group recognition By Vuong Le 3D face model from 2D face Audio-visual emotive avatar (driven by text and emotion markers) 3D face tracking and emotion recognition
Li Le Show
Recommend
More recommend