Despite the technological advances of security systems in recent times, we are unable to detect when a potentially dangerous person (such as a wanted criminal or any other blacklisted person) is within our vicinity. We wanted to take advantage of a security system already in place in many public areas to help safeguard these spaces against these threats.
What it does
Trackr automatically scans security camera footage for faces and compares them to our database of faces of wanted persons. It then automatically sends a text message to a point of contact (ex. a manager, police station, etc.) about the incident details along with a photo.
How I built it
We used the tracking.js library in order to detect and track faces in camera footage and screenshotted the video whenever we detected a new face. We then used the Microsoft Azure's facial detection API in order to compare these images against our database of wanted faces.
Challenges I ran into
We had difficulties integrating the frontend and backend, as well as tracking the faces on the webcam.
Accomplishments that I'm proud of
This was the first time all of us had worked with computer vision. In addition, it was the first hackathon for one of our members.
What I learned
We learned how to implement Microsoft Azure's facial identification API and the tracking.js library.
What's next for Trackr
We'd like to be able to expand this platform to missing persons and objects as well. For example, we could track the license plate numbers of stolen vehicles or people who have gone missing. In addition, the webcam feature is still not perfect and occasionally doesn't get the correct areas of the screen where our faces are. Also the webcam takes a screenshot every single time a new face is shown, which would produce a lot of unnecessary output. In the future we hope to be able to only take a picture when a match is found.