The problem E-SECUREity solves

Our solution obtains data from multiple CCTV sources and automatically identifies the vehicles being stolen, escaping criminals, and also the theft materials with the help of Computer Vision by providing it with the required data.

Our system automatically captures if any violent or criminal activity is taking place and it automatically alerts the nearby control center. We also have gesture controls where certain hand gestures can trigger the CCTV to indicate to the control center about any dispute in that region.

If any criminal activity is detected in any of the CCTVs, it will give an alert to the nearby CCTVs to look for further criminal activities in that area. So if any suspicious person involved in violent activities is found, the CCTV will automatically take their images and store them for further use and send them to the control centers.

Using the data from the CCTV cameras, if there are frequent criminal activities taking place in a particular region, the police department can easily mark it as a not safe or (risk zone ) and alert the citizens about it.

Additionally, our system can also detect road accidents and fire accidents. After analyzing the severity of that accident, it will indicate the ambulance service or fire engines to come to that region. It will also give the details about the owner of the vehicle and indicate to the police about the car owner’s details by analyzing the number plate of that car.

If a person or vehicle or any other distinct object is labeled as lost or stolen, our system will keep looking for those objects. If they are found later, the police will be intimated regarding it with the location details and other useful information.

Our system is a future-proof system and we can easily integrate new modules based on the prevailing situations. Example Scenario: During a situation like the covid pandemic , our system can be equipped with a social distance detection module within minutes.

Challenges we ran into

Implementing the realtime criminal detection was not an easy task to build, As identifying a person from a cam feed is computationally expensive we were facing slow down issues, We resolved it by load balancing and wrote a multiprocessing code which reduced the time taken significantly, Secondly we were facing the problem of identifying the numberplate on accident vehicles, We trained a custom Automatic number plate detection and recogniton tool to extract information about the vechicle that involved in an accident