Real-Time Criminal Identification System Using CNN-Based Facial Recognition and Haar Classifier on Mobile Platforms

EasyChair Preprint 15863

Abstract

Criminal identification systems are crucial in modern law enforcement, enabling rapid and accurate suspect recognition. This research presents real-time face recognition using an automated surveillance camera. The proposed system consists of 4 steps. First, training of real-time images. Then, face detection using the Haar-classifier. The next step is matching the security camera footage with the trained real-time images—the final result of the comparison. There is a need for automated monitoring systems that can detect watchlists. This research demonstrates a robust system capable of real-time face detection. In OpenCV, face tracking is accomplished using Haar-like classifiers. The facial recognition system works well. The suggested system's ability to recognize several faces quickly is invaluable for rapid suspect searches. The government of India uses Aadhaar to keep track of its citizens. To investigate crimes, we can utilize this citizenship database to distinguish between citizens and foreigners. Our dataset consists of 2000 images for five persons; every person has 400 face images. The system's accuracy is 95.5 % after splitting the dataset to 80% for training and 20% for testing using the CNN model

Keyphrases: CNN, Criminal Identification, Facial Recognition, Haar Classifier, mobile platforms

@booklet{EasyChair:15863,
  author    = {Ahmed Salman and Wisam Ali and Basim Abdul Kareem and Ammar Abdulhassan and Amjed Abbas and Abbas Fadhil},
  title     = {Real-Time Criminal Identification System Using  CNN-Based Facial Recognition and Haar Classifier on Mobile Platforms},
  howpublished = {EasyChair Preprint 15863},
  year      = {EasyChair, 2025}}