With fifteen years’ research and development experience on computer vision and pattern recognition, we mainly work on two areas.

    Deep learning. Based on the enmerging deep neural neworks, we are exploring transfer learning, reinforcement learning, and metric learning techniques to learn visual features with enhanced robustness and generlization ability. These visual representations can largely faciliate the applications on image retrieval, classification, and clustering.
    Human Sensing. Human plays a central role in computer vision applications. We aim to develop advanced methods to recognize and understand human and their behaviors from image and video. Our work is motivated by applications in the fields of video surveilance, human health, biometrics and human-machine interface.