Naimul Mefraz Khan
Software Systems
Surveillance cameras are common in private homes to strengthen security. A useful feature of a surveillance camera would be automated facial recognition. This can strengthen both the security aspect (alert user via smartphone if someone unrecognized is at the door) and social aspect (preparing for unexpected guests) of a smart home.
The target is to create a facial recognition enabled camera that can update a homeowner in real-time on who is at the door through an accompanying smartphone app.
1. The smart camera will have a microcomputer that can capture images and video and send to a server for facial recognition.
2. The server processes the face and compares with existing database facial images.
3. If a match is found, the homeowner is alerted about the person's identity through the smartphone app.
4. If a match is NOT found, an image of the person's face is sent to the homeowner through the app.
5. App should have abilities to a.) fetch ID of person from a database b.) show person's face if not recognized c.) add new person to the database. ALL of these must be achievable through the app itself with a few clicks, not manually.
1. Raspberry Pi/NVIDIA Jetson Nano with a camera module.
2. Server can be a cloud back-end or a central computer.
3. Server and camera module should be wirelessly connected, where the camera module sends data to the server.
4. Server will process the facial recognition (can use OpenCV/other free face recognition APIs).
1. Literature review (microcomputers, cloud, app building platforms, facial recognition methodologies)
2. System design
3. Implementation
4. Performance testing under different conditions (low lighting, different angles, poor connection, etc.)
Hardware
Programming for face recognition
Programming for image streaming from camera + backend development
App development (notification, adding new image to database, any other required customization)
NMK01: Smart Security Camera with Facial Recognition | Naimul Mefraz Khan | Sunday September 1st 2019 at 03:12 PM