Comparative Analysis of Classifiers for Criminal Identification System Using Face Recognition
Sanika Tanmay Ratnaparkhi, Pooja Singh, Aamani Tandasi, Nidhi Sindhwani · 2021
Computer vision and object detection software systems have evolved to solve problems that are hard to tackle. Many advances and research has been conducted in the field of object detection, facial tracking, face recognition etc. Security systems and biometrics remain an essential area where object detection and other techniques can be used. The main intention of this study is to develop an effective criminal identification system employing face detection and alignment using Multitask cascaded convolution neural networks, face embeddings using the application of “FaceNet” which harnesses the power of Convolution neural networks trained on a loss function and optimised using SGD method and finally a comparison of the performance of various Machine Learning classifiers to verify the identity of criminals. The dataset used here is the NIST mugshot dataset, which includes mugshots of offenders and lawbreakers. Tools such as precision, recall, accuracy and f1 scores are used to show the efficiency and working of the various models.