In this work, we propose a novel and efficient method for articulated humanpose estimation in videos using a convolutional network architecture, whichincorporates both color and motion features. We propose a new human body posedataset, FLIC-motion, that extends the FLIC dataset with additional motionfeatures. We apply our architecture to this dataset and report significantlybetter performance than current state-of-the-art pose detection systems.
展开▼
