We propose a novel video inpainting algorithm that simultaneously hallucinates missing appearance and motion (optical flow) information.
We introduce, TextureNet, a neural network architecture designed to extract features from high-resolution signals associated with 3D surface meshes (e.g., color texture maps).
We present a new, practical method that can reliably identify background features in complex video. leading to accurate camera path estimation and background layering.
We propose an image-based clothes changing system, exploiting body factor extraction and content-aware image warping.Paper
We developed algorithms for several tasks on TV news video, such as commercial detection, topic classification, host detection and clothing style classification etc. We applied these algorithms on a large scale TV news dataset, which contains more than 200000 hours of U.S. TV news.Paper (Coming soon)
We are aimed to determine the swimming motion pattern of an extinct species -- Eurypterid. Our final result provide evidence for supporting one of the swimming motion hypothesis.