Human-based Interaction Analysis via Automated Key point Detection and Neural Network Model

The human interaction with an object is one of the most challenging domains in real-life applications, such as smart homes, surveillance, medical, education, safety-based application computer vision, and artificial intelligence. In this research article, we have proposed a framework for examples sports other activities. Initially, reviewed video-based data by considering three state-of-the-art sets. Preprocessing steps been followed to avoid extra costs, video-to-frame conversion, noise reduction background subtraction. Human silhouette extraction has performed via Gaussian mixture model (GMM) supper pixel model. Next, body points location detection were performed. Finally, object-based features extracted. To minimize replication achieve optimized results, applied stochastic gradient descent Restricted Boltzmann Machine; As result, achieved accuracy 88.46%, 82.00%, 88.30% on parts recognition over MPII dataset, UCF_aerial wild Dataset respectively. classification dataset 92.71%, 90.60%, video 92.42%. We high rate compared methods frameworks due complex feature optimization approach.