The problem BrAIn solves

Surveillance is the monitoring of behavior, activities, or information to influence, manage or direct. Security surveillance is a major concerning issue with numeral news of breaches.

At present, a lot of our soldiers are busily patrolling the border areas cities, and compounds. There are various security systems in the market with CCTVs but they just record a scenario and all work of recognition is left to humans.

BRAIN: Camera with a Brain, is a smart surveillance system which uses facial recognition system with features like:

1. Creation of a Database of Users with their facial identities.
2. Real-time seamless Facial recognition from images, videos, CCTV streams.
3. Identify faces from the database, provide alerts when required.
4. Real-time alerts on suspected users(criminals), VIP with the location.
5. Track Suspect using Google maps.

The system is scalable with rapid deployment on the go with new faces just by adding a few images and details.

Platform and Accuracy:

1. Identifies faces typically up to a distance of 10 feet from the camera.
2. The actual distance depends on camera height and angle of incidence.
3. Input data can be RTSP(CCTV), HTTP video, image.
4. Our Facial recognition has accuracy over 99.5% on public standard datasets.
5. Recognition Can be used offline(without internet on edge devices).
6. Requires GPU, CPU based on no. of video channels and camera type.

Show Stopper
Real-time suspect notification and trackable location.

Challenges we ran into

We were testing our app in the drone but during the test, it crashed. Internet throttle was one of the challenges that we faced during the event. We were dealing with the hardware mostly so our initial time was spent on setting up the DVR with the CCTV camera. And since we brought the wrong monitor mistakenly we were unable to use it with our DVR, hance we chose move ahead. To make a real-life application the UI/UX should be great. But due to the higher computation that is required by the ML/Dl models, hence we optimized it to suit our requirements with trial and error.