 |
research |
|
news |
| This page is under heavy construction... In short, research in our lab concerns with high-level vision. That is, visual tasks involved in the semantic understanding of the visual world. My research interests involve both computer vision and human vision. In computer vision, we make algorithms and models to enable computers to perform visual tasks. In human vision, we conduct psychophysics experiments to uncover the underlying neuronal mechanisms that make humans see and interpret the visual scene. | ||
 |
Human action categorization In this project, we address the problem of classifying and localizing different human actions in video sequences. A video sequence is represented as a collection of spatial-temporal words by extracting space-time interest points. The algorithm learns the probability distributions of the spatial-temporal words and intermediate topics corresponding to human action categories automatically using a probabilistic Latent Semantic Analysis (pLSA) model. The learned model is then used for human action categorization and localization in a novel video. We also relax previous assumptions by allowing dynamic backgrounds and a moving camera. [Project page] |
|
