Home / Current Issue / Paper 1719977
FaceTrace: AI-Powered Real-Time Facial Recognition System for Integrated Home and Public Safety
Subject area: Science,Engineering and Technology · Area of research: Artificial Intelligence
DOI: https://doi.org/10.64388/IREV10I1-1719977
Abstract
The increasing demand for intelligent security sys-tems has highlighted the limitations of traditional surveillance solutions, which primarily rely on passive monitoring and lack real-time decision-making capabilities. In this paper, we present FaceTrace, an AI-powered real-time facial recognition system designed to enhance both residential and public safety through automated detection, classification, and alert generation. The proposed system integrates deep learning-based face detection using Multi-Task Cascaded Convolutional Neural Net-works (MTCNN) with feature extraction through the FaceNet architecture (InceptionResnetV1). Each detected face is trans-formed into a compact 128-dimensional embedding, enabling efficient and scalable identity recognition using distance-based matching. Unlike conventional classification-based models, the system employs an embedding-based incremental learning ap-proach, allowing new individuals to be added dynamically with-out retraining the model. FaceTrace follows a multi-module architecture comprising user, admin, and police station components. The system classifies individuals into four categories: familiar persons, unknown individuals, criminals, and missing persons. Based on the clas-sification, context-aware alerts are generated and categorized into normal, warning, and danger levels. A key feature of the system is its location-based alert routing mechanism, which automatically notifies the nearest police station in the event of high-risk detections. The system is implemented using a web-based interface for real-time video monitoring and a mobile application for user interaction, including profile management, alert history tracking, and complaint submission. Experimental evaluation demonstrates that the system achieves an overall accuracy of approximately 86% with low processing latency (approximately 1.2 seconds per frame) and efficient alert delivery (within 2 seconds). The confusion matrix analysis shows strong classification performance with minimal cross-category misclassification, ensuring reliable operation in practical scenarios. Overall, FaceTrace provides a scalable, efficient, and deploy-able solution that bridges the gap between personal security systems and public safety infrastructure. The integration of real-time facial recognition with intelligent alert mechanisms and law enforcement coordination makes the system a promising approach for next-generation smart surveillance applications.
Keywords
Facial Recognition, Deep Learning, FaceNet, MTCNN, Smart Surveillance, Real-Time Systems
How to cite this paper
@article{1719977,
author = {Kiran Krishnan V, Frijo Antony C F, Sandra Rose Joseph, Sharun K S, Dr. Joycy K Antony},
title = {FaceTrace: AI-Powered Real-Time Facial Recognition System for Integrated Home and Public Safety},
journal = {Iconic Research And Engineering Journals},
year = {2026},
volume = {10},
number = {1},
pages = {2612-2624},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1719977.pdf},
abstract = {The increasing demand for intelligent security sys-tems has highlighted the limitations of traditional surveillance solutions, which primarily rely on passive monitoring and lack real-time decision-making capabilities. In this paper, we present FaceTrace, an AI-powered real-time facial recognition system designed to enhance both residential and public safety through automated detection, classification, and alert generation. The proposed system integrates deep learning-based face detection using Multi-Task Cascaded Convolutional Neural Net-works (MTCNN) with feature extraction through the FaceNet architecture (InceptionResnetV1). Each detected face is trans-formed into a compact 128-dimensional embedding, enabling efficient and scalable identity recognition using distance-based matching. Unlike conventional classification-based models, the system employs an embedding-based incremental learning ap-proach, allowing new individuals to be added dynamically with-out retraining the model. FaceTrace follows a multi-module architecture comprising user, admin, and police station components. The system classifies individuals into four categories: familiar persons, unknown individuals, criminals, and missing persons. Based on the clas-sification, context-aware alerts are generated and categorized into normal, warning, and danger levels. A key feature of the system is its location-based alert routing mechanism, which automatically notifies the nearest police station in the event of high-risk detections. The system is implemented using a web-based interface for real-time video monitoring and a mobile application for user interaction, including profile management, alert history tracking, and complaint submission. Experimental evaluation demonstrates that the system achieves an overall accuracy of approximately 86% with low processing latency (approximately 1.2 seconds per frame) and efficient alert delivery (within 2 seconds). The confusion matrix analysis shows strong classification performance with minimal cross-category misclassification, ensuring reliable operation in practical scenarios.
Overall, FaceTrace provides a scalable, efficient, and deploy-able solution that bridges the gap between personal security systems and public safety infrastructure. The integration of real-time facial recognition with intelligent alert mechanisms and law enforcement coordination makes the system a promising approach for next-generation smart surveillance applications.},
keywords = {Facial Recognition, Deep Learning, FaceNet, MTCNN, Smart Surveillance, Real-Time Systems},
month = {July},
doi = {https://doi.org/10.64388/IREV10I1-1719977}
}