Currently, I am a Computer Vision Reseacher at Huawei. I recently graduated from M.Sc. of Computer Science, University of British Columbia, where I was working with Jim Little and Helge Rhodin.
My research interests are Computer Vision, Machine Learning and Natural Language Processing.
View My LinkedIn Profile
Project description: In this project, given a source video of a person dancing, we want to transfer the dance motion to an amateur target performing some standard moves. There are lots of applications for this work in advertisement or film industry. We approach this problem as a video to video translation using pose as an intermediate representation. To transfer the motion, we extract poses from the source subject and apply the learned pose to appearance mapping to generate the target subject. For the background, as opposed to the baseline model we use the source video, due to having more realistic features.
For more details see report.