Human Pose Estimation and Object Interaction for Sports Behaviour

In the new era of technology, daily human activities are becoming more challenging in terms monitoring complex scenes and backgrounds. To understand from life logs, human-object interaction (HOI) is important visual relationship detection pose estimation. Activities understanding recognition between object along with estimation modeling have been explained. Some existing algorithms feature extraction procedures complicated including accurate rare postures, occluded regions, unsatisfactory objects, especially small-sized objects. The HOI techniques instance-centric (object-based) where predicted all pairs. Such depends on appearance features spatial information. Therefore, we propose a novel approach to demonstrate that alone not sufficient predict HOI. Furthermore, detect body parts by using Gaussian Matric Model (GMM) followed YOLO. We points which directly classify pair them densely vectors algorithm. interactions linked actions. experiments performed two benchmark datasets demonstrating proposed approach.