Basic design for the implementation of automatic surveillance system on helmet detection
Mogalraj Kushal Dath, Manik Rakhra, Dalwinder Singh, Arun Singh, Rajesh Banala
Abstract
Deep learning has lately received acclaim for its success in certain fields, such as digital image pattern recognition and feature extraction. These methods have been used by researchers to solve a variety of issues, such as the detection of traffic violations specifically for motorcycle riders who are not wearing helmets in video surveillance. In this paper, we proposed a basic implementation and design steps of detecting two-wheeler bike rider wearing helmet utilizing a compatible and faster deep learning approach known as Single Shot Detector (SSD) in Linux operating system. We used and created a customized dataset of images by taking screenshots of a surveillance video of CCTV from a legal source. The traffic police can use this system to monitor the vehicles passing through specific surveillance nodes. After further implementation, number plates of vehicles could automatically be logged in a database which may be helpful in narrowing down options during a crime investigation.