- Home
- Archives
- 2026
- Vol. 6 Issue 6
- Article
Smart Border Intrusion Detection System (SBIDS)
Authors
1 Professor, Department of Information Science and Engineering, Bapuji Institute of Engineering and Technology, Davangere, Karnataka, India. 2 3 4 5 Department of Information Science and Engineering, Bapuji Institute of Engineering and Technology, Davangere, Karnataka, India.
Article Information
- Published In
- Volume 6, Issue 6 (2026)
- Published Online
- November-December 2026
- Pages
- 171-180
- Article Type
- Original Article
Abstract
View PDFAutomated perimeter surveillance faces challenges involving false alarms at restricted-zone boundaries and processing delays on resource-constrained hardware. This paper presents the Smart Border Intrusion Detection System (SBIDS), a multi-threaded computer vision prototype for restricted-zone monitoring. The system combines YOLO11n object detection, greedy tracking using Intersection-over-Union (IoU) and normalized foot-point distance, and a six-state finite state machine with a 5.0-second dwell-confirmation threshold. A dedicated latest-frame buffer separates video capture from inference and overwrites unread frames to limit application-level queue accumulation. In a controlled geometric simulation of 36 synthetic boundary positions, labelled by foot position, bounding-box intersection triggered on 12 of 21 overhang cases, whereas foot-point containment triggered on none. This illustrates the geometric origin of overhang false alarms; it is not an accuracy estimate for real scenes. In a 50-image CPU benchmark on an Intel Core i5-12450HX, the prototype recorded a mean end-to-end processing latency of 158.94 ± 27.95 ms, corresponding to approximately 6.29 FPS, below the configured target of 12 FPS. Historical telemetry from 193 developer-run operational events is used to characterize event statistics and system behaviour, not to establish detection accuracy. Recorded operational examples illustrate person and vehicle event handling. The system also supports immediate plain-text email alerts followed by asynchronous Vision-Language Model (VLM) incident summaries. The study discusses implementation limitations and deployment considerations.
Related Articles
Share Article
*Instagram doesn't support direct link sharing from web. Copy the link and share it in your Instagram story or post.