This Open Source Security Camera Tracks People with Cloud Computer Vision

Security cameras have come a long way since the grainy, black-and-white CCTV (Closed-Circuit Television) cameras that were originally introduced decades ago. Now you can purchase commercial security cameras that have fantastic picture quality, pan and tilt functionality, and even advanced features like object detection or people-tracking. However, few of those high-tech cameras are affordable, and even fewer of them are open source. FaceAssure is a cloud-based pan-tilt security camera that is completely open source so you can make it yourself.

A basic pan-tilt security camera will cost you at least $50, and you still need something to connect it to in order to utilize it. Once you’ve done that, all you can do is view the video feed and move the camera. That’s all well and good if you feel like constantly monitoring the feed, but doesn’t help much if you’ve got better things to do. FaceAssure was purpose built from the ground-up, and is smart enough to be able to track people if they enter the camera’s viewing area.

FaceAssure was built by putting a 2 megapixel camera sensor on a custom PCB. A Microchip ATSAM4S8BA microcontroller on that PCB grabs the video feed and controls the pan and tilt servo motors. The feed is then passed on to a cloud-based Tornado server via a built-in WiFi module. That server is running OpenCV computer vision software, which handles the detection and tracking of individuals. If a person moves across the frame, command signals will automatically be sent back to pan or tilt the camera as necessary. Users can view the video feed through a web interface. If you’ve got the skill to make this camera, it’s an option that is both affordable and capable.