About the Research Project
\
The Perception and Activity Understanding group seeks highly motivated PhD students to join a project focused on developing advanced AI models for real-world social behavior understanding. The project aims to capture the broader phenomenological aspects of where and why people look, integrating gaze with other social cues in natural, dynamic environments.
\
This contrasts with previous studies which had modeled gaze in tandem with other behavioral cues only in restricted settings. Leveraging cutting-edge deep learning methods, including recent contributions from our team, the project will analyze complex social interactions in daily scenes.
\
Main research areas include: 1) unified deep learning for head, facial, and upper body behavior recognition; 2) models for attention states and social relations analysis; and 3) machine learning strategies such as multi-task learning, unsupervised learning, and distillation to unify the modeling of these social behaviors.
\
Your Role in the Project
\
You will work on key tasks such as dataset creation, model design, and benchmarking, addressing challenges in diverse, unstructured settings. Successful candidates will hold an MS degree in computer science, engineering, physics, or applied mathematics.
\
A solid foundation in statistics, linear algebra, signal processing, machine learning, and programming is essential, with mandatory experience in deep learning. You will demonstrate strong analytical abilities and excellent written and oral communication skills.