ICCV’2019 HBU workshop, in addition to covering the main themes of human behavior understanding, deals with generating human behavior data, with special emphasis on methodologies and approaches for forging images and videos depicting real-looking human faces and/or full bodies and on algorithms for detecting fake human-like visual data. Contributions based on deep neural architectures are welcome, as well as methods based on other techniques (e.g. parametric models). These contributions could address the following topics:
Human Behavior Analysis Systems
- Action and activity recognition
- Affect analysis
- Face analysis
- Gaze, attention and saliency
- Gestures and haptic interaction
- Social signal processing
- Voice and speech analysis
- Theoretical frameworks of behavior analysis
- Data collection, annotation, and benchmarking
- User studies and human factors
Generating Visual data of Human Behavior
- Methods for face synthesis and modification of facial attributes (e.g. age, expression).
- Approaches for generating human bodies and altering their properties (e.g. 3D pose, clothes).
- Techniques for forging human-like behavioral data
- Methodologies for counteracting adversarial attacks.
- Techniques for synthesizing visual data depicting collective human behaviour.
- Novel deep generative models for sequence-like data generation.
- Approaches to synthesize multi-modal human behavioral data.
- Applications (e.g. surveillance, entertainment, autonomous driving, fashion, robotics).