上一条:Multi-Channel Co-Attention Network for Visual Question Answering
下一条:Multi-template supervised descent method for face alignment