%D 2024
%R 10.1109/KST61284.2024.10499688
%O cited By 0; Conference of 16th International Conference on Knowledge and Smart Technology, KST 2024 ; Conference Date: 28 February 2024 Through 2 March 2024; Conference Code:198941
%L scholars20037
%J KST 2024 - 16th International Conference on Knowledge and Smart Technology
%K Alarm systems; Automation; Deep learning; Intelligent buildings; Internet of things; Learning algorithms, Curated dataset; Daily lives; Environmental information; Environmental sensing; Healthcare systems; Information collections; Intelligent transportation; Intrusion dataset; Intrusion-Detection; Perimeter intrusion detection dataset, Intrusion detection
%X The Internet of Things (IoTs) has become ubiquitous in daily life, contributing to environmental sensing and information collection in areas such as smart cities, healthcare systems, intelligent transportation, and smart homes. Additionally, an important application of IoT is Perimeter Intrusion Detection (PID), crucial for monitoring commercial, residential, or industrial locations. While many of the researchers have been focused on intrusion detection systems but PID still lacking a particular dataset for experimenting with machine learning and deep learning algorithms to find out the better false alarm rate (FAR) and probability of detection (POD). Considering these challenges, this paper introduces a curated dataset named ST-PID, captured from various cameras at different locations, recording videos over a three-month period, including day and night monitoring. The dataset is enriched through preprocessing, image augmentation, and labeling, resulting in 8000 images depicting various intrusion activities. YOLO V5 is then applied to successfully detect six classes of intrusion activities for dataset evaluation. © 2024 IEEE.
%P 85-90
%T Curated Dataset Generation for Enhanced Security: ST-PID Dataset
%A S. Pitafi
%A T. Anwar
%A Z. Sharif
%A Hina
%A Z. Hassan