OCULAR: Object detecting CCTV using a low-cost artificial intelligence system with real-time analysis

Abstract

The essential aspect of a country's self-sustenance is security. With the increasing number of vehicles per year, and as a vehicle enters a gated area or community, security personnel cannot properly handle the number of license plates to inspect and manually record in the outpost. Vehicle plate recording is still being done manually in most institution. Without a school sticker, the security personnel encode the license plate of an incoming vehicle and only gives the parking ticket or visitor's pass to the driver. The researchers were able to develop an Object Detecting CCTV using Low-Cost Artificial Intelligence System with Real-Time Analysis to solve this particular problem. This research was developed through image processing and deep learning methods. Images of license plates were collected and used to train an object detection model, wherein license plates are localized, detected, and undergo Optical Character Recognition. OCULAR presents an automatic license plate recognition system that can detect, recognize, and record license plates while providing real-time video streaming and video footage. It has also an alert system to check and identify the status of the vehicle, if it is registered or not. All data detected and video footage recorded by the Raspberry Pi Camera v2 is stored in a SQLite database and can be viewed in a user interface. In addition to that, video footage from the camera is streamed to the main server and will be stored on the server's hard drive. Through Artificial Intelligence, 99.89% of license plates and 91.02% of the license plate numbers were detected and recognized properly and precisely. The system has an overall accuracy of 95.33%. Therefore, the device and the system were highly reliable for the proper detection and recording of vehicular license plates.