Fall detection with the Livy Care System

To enable older individuals to live independently in their own homes for as long as possible while still being able to react quickly in case of emergency, technologies are needed that automatically detect when someone has fallen and needs help. The Livy Care System by HUM Systems GmbH is one of these technologies, which uses radar technology and computer vision to detect when a person is on the ground and directly notifies relatives or caregivers from the care service.

To prevent a person lying in bed from being recognized as “fallen,” the bed must be marked as a so-called “safe zone.” So far, the “safe zone” has been manually marked when setting up the device. However, when using the Livy Care System in a stationary nursing home or hospital, there is a special consideration that the bed is often moved. To avoid having to reset the safe zone every time the bed is moved, HUM Systems GmbH wants to use an AI-based algorithm to automatically recognize the bed and its position.

To train this algorithm, HUM Systems GmbH needed several hundred images of different beds in different rooms, positions, and lighting conditions. These images were captured at the European Digital Innovation Center CITAH in the home and stationary care laboratory of OFFIS e.V. Subsequently, the CITAH team labeled the captured images, meaning the elements recognizable in the image were manually labeled. Based on the labeled images, the automatic detection of the “safe zone” could now be realized. In the further course of the project, it is planned to make these training data available as an openly accessible dataset for other companies and use cases.