A KEY POSE MODEL FOR HUMAN INTERACTION RECOGNITION AND COLOR FROM GRAY BY OPTIMIZED COLOR ORDERING

摘要: In this thesis, we address two different problems in computer vision. First, the human interaction recognition is discussed whose goal to recognize action type of interacting humans videos. We model using a sequence key poses, important atomic-level actions. Spatial arrangements between actors are included as strict temporal ordering poses. Second, attack problem color from gray. Our build gray level image which can be used recover with minimum amount embedded information. use parametric curve maps values rich samples space. Next, attach parameters for order an approximation color. The method rests on optimization both and error minimized.