A Modular AI-Powered CCTV Surveillance Architecture for Action Recognition and Decision Support

View at publisher Summarise with VIRAT
Authors: M. Sathiya, S Girija Lakshmi, Bala Murgan S, Mounish S, Soundarya S, Ziya Sherzeen S
Cited by: 0
References: 15

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents a comprehensive AI-powered intelligent CCTV surveillance system integrating real-time computer vision, multi-object tracking, rule-based safety monitoring, and natural language query capabilities. The proposed system comprises four integrated subsystems: Vision (YOLOv11 detection, pose estimation, action recognition, attribute detection), Tracking (ByteTrack, Re-ID embedding, Kalman filtering), Intelligence (polygon-zone monitoring, JSON rules engine, automated alert dispatch), and Chatbot (intent classification, database queries, evidence retrieval). The system achieves 95%+ overall detection accuracy, 77.3% MOTA tracking performance, and processes surveillance feeds at 30 frames per second with 35-40ms end-to-end latency. Experimental validation demonstrates the system's capability to answer natural language queries such as "Did I lock the doorƒ" within 3 seconds while providing video evidence with 96% confidence. The architecture supports real-time safety compliance monitoring in industrial environments, automated threat detection in security applications, and instant evidence retrieval for household scenarios. Performance evaluation on standard benchmarks confirms the system's suitability for deployment on RTX 3080-class hardware supporting 4-camera configurations. The integrated approach reduces manual video review time from 15 minutes to 3 seconds and enables proactive safety interventions through automated alerts with severity-based escalation protocols.

Fields of Study & Subjects

Video Surveillance and Tracking Methods Human Pose and Action Recognition Advanced Neural Network Applications

Publication details

Journal / source
International Conference on Image Processing and Capsule Networks (ICIPCN)
Pages
502–507
Publication date
27 January 2026
Language
EN