Tools
Tools
Jabarullah, Mahir Ahmed (2024) Deep-learning and Cloud-based IoT framework for intrusion detection using video surveillance. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (1MB) | Preview |
Preview |
PDF (Configuration Manual)
Download (1MB) | Preview |
Abstract
Traditional intrusion detection systems suffer from the delays and inefficiencies, due to the rule-based detection or manual monitoring. By employing scalable cloud services and lightweight edge computing, the proposed system enhances detection accuracy, speed, and resource efficiency. This research bridges the gap between robust detection systems and practical applications by integrating state-of-the-art models like FaceNet and InceptionResNet with HAAR cascades and SVMs for edge deployment. Challenges such as illumination variations and computational constraints on edge devices were addressed through quantization and advanced preprocessing techniques. The Key results demonstrate a detection accuracy of 97 % in live scenarios and scalability in cloud environments compared to the YOLO and SORT models, which has been used in the traditional system. This comprehensive framework ensures dynamic, cost-effective and secure monitoring for residential and commercial settings, marking a significant step toward smarter surveillance systems.
Actions (login required)
 |
View Item |