Applied Depthwise Separable-based Convolutional Neural Network Classification on MobileNet Algorithm for Human Fall Detection
Pipat Sakarin, Suchada Sitjongsataporn · 2024
This paper presents the human fall detection by convolutional neural network (CNN) classification and MobileNet algorithm. In order to reduce the number of parameter, we propose a vision-based fall detection model, that is called “Applied Depthwise separable-based CNN-MobileNet (ACM) algorithm”. This model is based on CNN, deptwise convolution and pointwise convolution of MobileNet algorithm. Experimental results show the accuracy of proposed ACM model about 97.68% and number of parameters of proposed ACM model is less than conventional CNN model.