We present a way to deal with effectively distinguish the 2D posture of different individuals in a picture. The methodology utilizes nonparametric portrayal, which we imply to as Part Affinity Fields (PAFs) which figure out how it partners body parts with people in the photo. This engineering utilizes worldwide setting, permitting an eager base up parsing step that keeps up high precision while accomplishing real-time execution, independent of the quantity of individuals in the picture. The engineering is intended to together adapt part areas and their affiliation by means of two parts of the equivalent successive expectation process.