Agent Behavior Understanding in Crowds – Predicting Future Trajectories and Activities


At a glance

Agent behavior understanding in crowds is deemed a rather complicated task due to the variability of motion flows undertaken by different individuals, especially in dense settings where the intricacy multiplies. Precisely, motion trends of an individual are consistent with and often dependent upon his/her spatial context in a given dense crowd setting. This suggests that understanding the behavior (e.g., walking direction) of a person can be modeled by studying his/her context. In this proposal, our target is to provide an end-to-end, multi-task learning system utilizing rich visual features about human behavioral information and interaction with their surroundings. A network is trained to predict future activities and the location where they will happen from large scale camera data of human interactions. Previous work on crowd behavior analysis, future forecasting, activity recognition and unsupervised embedding of camera images will be leveraged.


Ching-Yao Chan
Stella Yu

Sascha Hornauer
Yuke Li

Multi-Agent, Multi-Camera, Pedestrian Action Prediction, Behaviors in Crowd, Human Interaction with Surrounding, Multi-Task Learning