Hybrid CNN models for suspect facial recognition system

The criminal landscape is evolving rapidly. Perpetrators are constantly upping their technological game, exploiting new tools for nefarious purposes. Unfortunately, law enforcement isn’t always keeping pace. While a seasoned deceiver might be adept at hiding their true colors, there are often involuntary tells – especially in facial expressions. However, pinpointing these subtle clues in a bustling environment can be a real headache for authorities. Face recognition is a vital application in computer vision with many real-world use cases, such as security systems, access control, and authentication. It is worth mentioning that the Convolution Neural Network (CNN) plays an important role in speeding up and enhancing the accuracy of facial recognition algorithms to assist in surveillance systems and access control. This project is designed to empower various authorities, including government agencies, international airports, and Egyptian border security, to enhance their investigative capabilities. by using more efficient models for face detection and recognition to build a hybrid algorithm of two CNN models to detect and predict suspicious persons and report to the authorities. Beyond facial recognition capabilities, the proposed model incorporates face detection functionalities. This will enable the identification of individuals flagged for travel restrictions or suspected of attempting unauthorized departures. This paper presents a survey of popular face recognition models using CNN and explains their features. In addition to the survey, it integrates two algorithms for detection and recognition based on a detailed analysis applied to each presented model, explaining their accuracy and the most crucial factor, processing time. The paper also introduces a practical use case that can be established in any crowded region.