Computer vision based fall detection by a convolutional neural network

<p>In this work, we propose a novel computer vision based fall detectionsystem, which could be applied for the health-care of theelderly people community. For a recorded video stream, backgroundsubtraction is firstly applied to extract the human body silhouette.Extracted silhouettes corresponding to daily activities are appliedto construct a convolutional neural network, which is applied forclassification of different classes of human postures (e.g., bend,stand, lie and sit) and detection of a fall event (i.e., lying postureis detected in the floor region). As far as we know, this work isthe first attempt for the application of the convolutional neuralnetwork for the fall detection application. From a dataset of dailyactivities recorded from multiple people, we show that the proposedmethod both achieves higher postures classification results thanthe state-of-the-art classifiers and can successfully detect the fallevent with a low false alarm rate.</p>