A Computational Model to Analyze Human Motion Identification Through Gait Analysis Using CNN

The wearable device and kinetic sensor-oriented human activity identification approach has recognized seven human activity using Convolutional Neural Network (CNN) technique. This easily available wearable Smartphone that has accelerometer and gyro meter for the purpose of data recording in terms of volunteer activities on ideal and zigzag surfaces. For ideal surface we have utilized a kinetic sensor device to sense the related data. The hardware devices such as wearable and kinetic sensors are capturing data in x, y, and z axis co-ordinates through accelerometer and gyro meter. After preprocessing of the primary dataset, feature extraction and classification process has performed followed by training, testing, and validating it by proposed CNN deep learning algorithm. In this paper, we classified seven human activities with the help of GDOHA dataset; with this model, we have achieved 99.07% accuracy, 96% precision, and recall of 97%. Novelty of this research paper is to increase the classification accuracy and reduce the time complexity of existing model.