ABSTRACT
In India, there are various factors, including an exponentially increasing population, an increased economic crisis, and a job recession, that have collectively led to an increase in the overall crime rate. Timely detection of felonies and immediate identification of felons is a crucial step in bringing the crime rate down. This must be the initial and major objective of any investigative authority. The emergence of Industry 4.0, the Internet of Things (IoT) and Machine Learning are all gaining momentum and importance in all fields, especially in the establishment of Smart Cities. In Israel, several Smart City initiatives and projects are being carried out to tackle major challenges in transportation, disaster management, security, etc. In the case of security, a number of steps and strategies have been introduced, including connecting to video feeds and automatically detecting the presence of weapons in crowded spaces, summarizing full-length videos and reducing the burden involved in browsing through long videos, installation of CCTV surveillance cameras at strategic points, etc.
In this chapter, we present an IoT and Machine Learning–based criminal identification system, using face recognition and text summarization, which takes images from surveillance cameras as input. The application makes use of a Local Binary Pattern (LBP) algorithm for face recognition and a TextRank algorithm for criminal record summarization. This system can be used by any institution or organization that attempts to identify lawbreakers who ardently wait for opportunities to commit felonies in crowded areas such as malls and airports. The system is made to learn various features from a database of criminal facial images, which typically contains images pertaining to different views of a single criminal. The summary along with the predicted label of the criminal would be helpful to private investigators or police forces in taking immediate preventive measures.
