Using Convolutional Neural Network for Human Posture Estimation: A study of the effects of number of layers and number of neurons on accuracy

Abstract

Human Posture estimation is a field which gathers huge researchers interest due to its variations in different machine learning (ML) & deep learning (DL) architectures to estimate human postures This work includes tweaking of layers with varying neurons in Convolutional Neural Network Architecture to test which pair of neurons and layers gives the best accuracy, also visualizing each of the pair with help of graphs. The results of this work provide great results with high accuracy.