Tools
Tools
Kanniah, Vijayakumar (2023) Connected Home Security System – An AI enabled surveillance system to detecting Unwanted Intrusions using Cloud Based system. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (730kB) | Preview |
Preview |
PDF (Configuration manual)
Download (1MB) | Preview |
Abstract
This paper presents a comprehensive facial recognition system that explores various deep learning and transfer learning models. Among them, the final model selection is based on three distinct approaches: Haar cascade, AdaBoost, and HOG+SVM. The selection criteria involve evaluating space consumption, memory utilization, and detection speed. This systematic model testing ensures an efficient and practical solution for real-time face detection and recognition. A key novelty of this system lies in its ability to not only identify known individuals but also detect and handle probable unknown people effectively. This feature enhances the system's security capabilities, protecting the house from potential attacks or unauthorized access attempts. By combining multiple detection methods, the system achieves robust and accurate identification of individuals, bolstering home security and surveillance. The research also leverages the advantages of deep learning and transfer learning techniques to optimize the models' performance. The training modules deployed on the AWS platform enable the system to continuously learn and adapt, generating the best model for each user over time. Overall, this paper showcases the innovative use of diverse approaches in facial recognition technology, considering both efficiency and security aspects. The proposed system not only excels in identifying known individuals but also effectively addresses the challenge of handling probable unknown people, making it a promising solution for improving home security and surveillance in a connected and intelligent environment.
Actions (login required)
 |
View Item |