My Research in RPI
1. Activity and Event Recognition
We focus on activity and event recognition in videos such as VIRAT. Contextual information from feature, semantic, and prior levels can be incorporated to improve the recognition performance.
2. Object Recognition with Attributes
We explore to automatically discover and capture the semantic relationships between attributes, and utilize the captured relationships to benefit both attribute prediction and object recognition.
3. Probabilistic Graphical Models
Probabilistic Graphical Models (PGMs) are our tools for various tasks. The research on PGMS is both challenging and interesting!