Human Pose Detection And Classification Using Movenet With Spatiotemporal Data

The evolution of human pose detection has seen significant growth with the development of numerous algorithms and the availability of vast datasets. Convolutional Neural Networks (CNN) algorithms offer diverse insights and prediction methodologies, contributing to the precision of pose detection rates. In this study, the focus is on analyzing the performance of the MoveNet algorithm in human pose detection. The MoveNet algorithm excels in extracting human pose keypoints from 17 locations on the body, including ears, eyes, nose, shoulders, elbows, wrists, knees, and ankles. Utilizing these detected keypoints, the algorithm estimates the human pose by generating predictions on the images.To enhance the computational efficiency of the model, the MoveNet algorithm is integrated with spatiotemporal data. The incorporation of spatiotemporal data elevates model accuracy by providing related images of the same event. Subsequently, the performance of the model is evaluated using standard metrics. This innovative approach opens new frontiers for exploring the relationships between human actions and their underlying intentions. The study contributes to the field by introducing a novel methodology that combines the MoveNet algorithm with spatiotemporal data, offering a promising avenue for improved human pose detection.