Border Security System Using Real-Time Image Processing

Abstract

In recent times, it has occurred to us that county borders are now more deadly with security flaws. There are millions of stuff being smuggled in and out, neighbour countries security forces having many ways to advance with gunning down people and entering the border area, seizing people and extortion. That's why it’s our effort to detect forces with a faster image processing method so the security personnel can use it to their benefit. Object detection in modern computer science has developed quite a lot in recent years. With the development of neural network algorithms, some notable of them are CNN, RCNN, and faster-RCNN algorithms. Our motive was to develop a real-time object detection system which can be used at the country border to help detect targets in order to increase security. That's why we used the YOLO (you only look once) algorithm to train and test out data. The YOLO algorithm takes a different approach in order to detect objects faster. In this research project, we trained raw data in YOLO and SSD (Single shot multibox detector) and compared their advantages and disadvantages for having a real-time level of detection and accuracy. This detection scheme can be applied in surveillance systems such as cameras, drones and video surveillance, which will require cloud and server-based processing in object detection Application Programming Interface. we’re hopeful that This research project could be one of the early steps to increase border area security.