AI-Enabled Sensor-Based Human Activity Recognition for Automated Surveillance

A Raspberry Pi microcontroller and a mobile application for real-time tracking and reporting of specified human activities were chosen in light of the activities' vital importance. In order to provide immediate rescue assistance, aged and sick individuals were observed using a camera installed to cross-communicate changes in their postures on the bed over a wireless network. Because they lack intelligent real-time capturing and reporting technologies, the conventional methods of keeping an eye on the aforementioned individuals using cameras or other people prove to be ineffectual. Notifications are sent to the mobile application by the video recorders. Modern computer vision and artificial intelligence techniques, which enable highly accurate human posture detection and recognition, fuelled the system's intelligence. To extract hidden information from photographs, the system was trained using Convolutional Neural Networks on the Common Objects in the Context dataset. This system uses mobileNet, Posenet, and Single Shot Detection to track human positions in real time and estimate postures.