Automatic Fall Detection System with a RGB-D Camera using a Hidden Markov Model

Abstract

Abstract. Falls in the elderly is a major public health problem because of their frequency and their medical and social consequences. New smart assistive technologies and Health Telematics make it possible to provide elderly with more security and well being at home. A smart home can automatically monitor home activities for early warning in health changes or detecting dangerous situations. One of our objectives is to design an automatic system to detect fall at home, which in its final version will be made up of a network of RGB-D sensors. In this paper, we present a simple and robust method based on the identification and tracking of the center of mass of people evolving in an indoor environment. Using a simple Hidden Markov Model whose observations are the position of the center of mass, its velocity and the general shape of the body, we can surprisingly monitor the activity of a person with high accuracy and thus detect falls with very good accuracy without false positives. An experimental study, that is reported here, has been driven in our smart apartment lab. 26 subjects were asked to perform a predefined scenario in which they realized a set of eight postures. 2 hours of video (216 000 frames) were recorded for the evaluation, half of it being used for the training of the model. The system detected the falls without false positives. This result encourages us to use this system in real situation for a better study of its efficiency.

HMM-Based Action Recognition System for Elderly Healthcare by Colorizing Depth Map

“…For systems relying on vision sensor data, the studies in [36,37] proposed HMMbased automatic fall detection systems with image processing techniques by utilizing RGB and RGB-D cameras, respectively. In [36], HMM was used as a decision-making process for differentiating abnormal (falling) from normal sequential states for a given person.…”

“…The HMM model was then developed by defining feature thresholds and calculating emission probabilities. On the other hand, [37] created an HMM model to detect and distinguish falling events from the other eight activities of the person. The observation symbols of their model were the vertical position of the center of mass, the vertical speed, and the standard deviation of all the points belonging to the person.…”

Non-overlapping Dual Camera Fall Detection Using the Nao Humanoid Robot

“…Of the most recent related literature the work generally falls into two categories, those that use RGB-D data taken from sensors such as the Microsoft Kinect [3] [4] and those that rely on RGB data alone. While RGB-D approaches generally provide very accurate fall detection and appear to be a promising line of research, computational requirements and issues with IR sunlight saturation remain a prohibitive factor for widespread uses.…”