Deep Learning for Computer Vision

Deep learning for computer vision aims at learning image presentations from large collections of visual data. Leveraged by the powerful Graphical Processing Unit (GPU), deep learning has become one of the fastest advancing field in computer vision, and outperform many traditional techniques in a number of benchmark datasets in multiple areas.

Our colleagues have successfully organized the first Deep Learning Workshop in the IEEE Computer Vision and Pattern Recognition, the flagship conference for vision, and published many deep learning papers in top conferences. Our applications include road scene analysis, face analysis, image quality assessment, 2D/3D shape retrieval and recognition, robotics, action analysis.