Date of Graduation


Document Type


Degree Name

Master of Science in Computer Engineering (MSCmpE)

Degree Level



Computer Science & Computer Engineering


Christophe Bobda

Committee Member

Jia Di

Second Committee Member

John Gauch


As technology advances in the field of Computer Vision, new applications will emerge. One device that has emerged is the smart-camera, a camera attached to an embedded system that can perform routines a regular camera could not, such as object or event detection. In this thesis we describe a smart-camera system we designed, implemented, and evaluated for fall prevention monitoring of at-risk people while in bed, whether it be for a hospital patient, nursing home resident, or at home elderly resident. The camera will give a nurse or caregiver environmental awareness of the at-risk person and notify them when that person performs an action that could lead to a hazardous event. This camera uses Haar Cascade facial detection techniques, Histogram of Oriented Gradients(HOG) for person detection, and Mixture of Gaussians (MOG) background subtraction while operating. Regions are created by a person from a graphical user interface (GUI). The camera looks within these regions to find a face, a standing person, or just a change in the image. A notification is sent to the smartphone of the nurse or caregiver of the corresponding at-risk person when the camera finds one of these three detections in the drawn region. The Cloud is utilized to send the notification to the nurse or caregiver’s smartphone. Given a properly placed camera and properly drawn regions, notifications can be sent when the at-risk person is doing an action that demands the attention of the nurse or caregiver, such as getting out of bed. The smart-camera does contain drawbacks. It is likely to give alerts when visitors are in the room, and it does not know how to pause notifications when a nurse, doctor, or caregiver comes into the room.