Video-based Behavior Analysis
Event details
| Date | 04.06.2018 |
| Hour | 11:00 › 13:00 |
| Speaker | Stefano Savaré |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Martin Jaggi
Thesis advisor: Prof. Pascal Fua
Co-examiner: Dr. François Fleuret
Abstract
Modeling and analyzing people behavior in crowded scenes can greatly improve
computer algorithms for scene understanding.
In this report we review 3 different approaches used in this field.
The first is based on an advanced pedestrians motion model that keeps into account
human-human, human-crowds and human-scene interactions.
The second exploits Long Short-Term Memory networks (LSTM) to learn the causalities behind human
motion in a completely data-driven approach.
The third, based on the Lagrangian framework for fluid dynamics, performs crowds analysis
through a streakline representation of the flow.
Finally, we suggest a new line of research to perform behavior analysis for groups of people.
Background papers
Pedestrian Behavior Modeling From Stationary Crowds With Applications to Intelligent Surveillance, by Yi, S., et al.
Learning to predict human behaviour in crowded scenes, by Alahi A., et al.
A Streakline Representation of Flow in Crowded Scenes, by Mehran R., et al.
Exam president: Prof. Martin Jaggi
Thesis advisor: Prof. Pascal Fua
Co-examiner: Dr. François Fleuret
Abstract
Modeling and analyzing people behavior in crowded scenes can greatly improve
computer algorithms for scene understanding.
In this report we review 3 different approaches used in this field.
The first is based on an advanced pedestrians motion model that keeps into account
human-human, human-crowds and human-scene interactions.
The second exploits Long Short-Term Memory networks (LSTM) to learn the causalities behind human
motion in a completely data-driven approach.
The third, based on the Lagrangian framework for fluid dynamics, performs crowds analysis
through a streakline representation of the flow.
Finally, we suggest a new line of research to perform behavior analysis for groups of people.
Background papers
Pedestrian Behavior Modeling From Stationary Crowds With Applications to Intelligent Surveillance, by Yi, S., et al.
Learning to predict human behaviour in crowded scenes, by Alahi A., et al.
A Streakline Representation of Flow in Crowded Scenes, by Mehran R., et al.
Practical information
- General public
- Free
Contact
- EDIC - [email protected]