Convolutional neural networks (CNN) based human fall detection on Body Sensor Networks (BSN) sensor data

According to the World Health Organization, around 28-35% of people aged 65 and older fall each year. This number increases to around 32-42% for people over 70 years old. For this reason, this research targets the exploration of the role of Convolutional Neural Networks(CNN) in human fall detection. There are a number of current solutions related to fall detection; however, remain low detection accuracy. Although CNN has proven a powerful technique for image recognition problems, and the CNN library in Matlab was designed to work with either images or matrices, this research explored how to apply CNN to streaming sensor data, collected from Body Sensor Networks (BSN), in order to improve the fall detection accuracy. The idea of this research is that given the stream data sets as input, we converted them into images before applying CNN. The final accuracy result achieved is, to the best of our knowledge, the highest compared to other proposed methods: 92.3%.