Research on Multi-person pose estimation is partly improved by deep learning and the computer vision. Multi-person pose estimation is expected to be involved in many applications, such as fitness training, pedestrian recognition, military training, and so on. The prospect of multi-person estimation development is promising and challenging. This paper provides a brief survey on four major multi-person pose estimation methods - DeepCut, Dee-perCut, OpenPose and AlphaPose, and presents the advantages and disadvantages of these methods.
展开▼