The paper presents OpenPose, a real-time multi-person 2D pose estimation system that uses Part Affinity Fields (PAFs) to associate body parts with individuals in images. The method achieves high accuracy and real-time performance, regardless of the number of people in the image. The authors demonstrate that refining PAFs alone, rather than both PAFs and body part location refinement, significantly improves both runtime and accuracy. They also introduce the first combined body and foot keypoint detector, which reduces inference time and maintains accuracy. OpenPose is the first open-source real-time system for multi-person 2D pose detection, including body, foot, hand, and facial keypoints. The paper includes a detailed description of the method, network architecture, and evaluation on multiple benchmarks, showing superior performance compared to existing methods.The paper presents OpenPose, a real-time multi-person 2D pose estimation system that uses Part Affinity Fields (PAFs) to associate body parts with individuals in images. The method achieves high accuracy and real-time performance, regardless of the number of people in the image. The authors demonstrate that refining PAFs alone, rather than both PAFs and body part location refinement, significantly improves both runtime and accuracy. They also introduce the first combined body and foot keypoint detector, which reduces inference time and maintains accuracy. OpenPose is the first open-source real-time system for multi-person 2D pose detection, including body, foot, hand, and facial keypoints. The paper includes a detailed description of the method, network architecture, and evaluation on multiple benchmarks, showing superior performance compared to existing methods.