Multi-Person Human Pose Estimation In Real Time - Exercise Demo

Through radical computer vision techniques, the key points of a person or an object are tracked to predict and track behaviours over the time. In the example of a Pose estimation applied on a human body, localizes certain key body joints(key points) in images and videos. These spatial locations can be used to determine if a person is standing, sitting, lying down or performing activities like dancing, jumping, etc. The application draws points on detected joints and semantically links them.

In the wake of Covid 19, the concept of home fitness through online workout videos has really gone up in addition to the increased popularity of health apps for round the clock health monitoring. The ability to track human movement and estimate the position of the person working out becomes extremely handy through pose estimation. Body pose analysis is being used in the field of sports by identifying the movement of players to better manage the game. While the athletes or players are training, their body pose estimation can send important alerts such as level of fatigue, stress or injury prompting the game to be stopped.

Our lives are increasingly getting automated with the help of Robotic instruments around us. The biggest benefit has been in health care where Robots are already being used inside Operation theatres to perform complex tasks. And with the advancement in 3D pose estimation, the ability of Robots to perform tasks a lot more accurately and responsively has gone up. Not only this but interactive video gaming has been made possible by motion capture technology powered by Deep Learning based pose estimation models. Now character animation can be done by capturing motion including various poses in real time.