Weight Training Pose Estimations

The concept of correct posture and ‘form’ has always been integral to weight training, especially when dealing with relatively heavier weights. The concept of pose estimation for humans has been a widely explored avenue by researchers, but applying the same for calibration of muscle flexions is something not explored enough. This study touches upon the subject and explores a variety of approaches and methodologies proposed in recent years based on existing pose datasets comprising of both images and videos. The proposed methods range from application of Greedy Bipartite matching algorithms to OpenPose and DensePose etc. with varying degrees of recorded accuracies and throughputs. This study is an attempt to condense the processes and methods employed for human pose estimation and specific examples pertaining to weight training.