Sistem Deteksi Kekerasan Dan Senjata Pada Rekaman Video Berbasis CCTV Menggunakan YOLOv11

Januriansyah, Muhammad Syarif (2026) Sistem Deteksi Kekerasan Dan Senjata Pada Rekaman Video Berbasis CCTV Menggunakan YOLOv11. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6026232017-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Kekerasan fisik dan pencurian bersenjata merupakan ancaman serius terhadap keamanan masyarakat, termasuk di Indonesia. Peningkatan kasus kejahatan yang melibatkan penganiayaan, pengeroyokan, serta penyalahgunaan senjata tajam dan senjata api menuntut adanya sistem pengawasan yang lebih efektif dan responsif. Pengawasan manual berbasis CCTV memiliki keterbatasan, terutama akibat kelelahan petugas dan potensi kesalahan dalam mendeteksi ancaman secara real-time. Penelitian ini mengusulkan pengembangan sistem pengawasan berbasis CCTV menggunakan algoritma YOLOv11 yang mampu mendeteksi tindakan kekerasan dan keberadaan senjata secara bersamaan dalam satu sistem terintegrasi. Sistem dikembangkan dalam bentuk aplikasi desktop dan dilengkapi dengan mekanisme notifikasi Telegram sebagai early warning system untuk memberikan peringatan dini ketika terdeteksi ancaman yang valid. Hasil evaluasi menunjukkan bahwa model YOLOv11 memperoleh performa yang baik dengan nilai mean Average Precision (mAP) keseluruhan sebesar 0,914, serta nilai precision dan recall masing-masing sebesar 0,906 dan 0,868. Sistem mampu melakukan deteksi secara real-time dengan performa stabil pada kisaran 30–40 FPS, mendukung pemrosesan video secara langsung maupun melalui unggahan berkas, serta mengirimkan notifikasi Telegram dalam waktu kurang dari dua detik setelah terdeteksi ancaman. Hasil ini menunjukkan bahwa sistem yang dikembangkan dapat berfungsi sebagai sistem pendukung pengawasan keamanan berbasis video.
===================================================================================================================================================================================
Physical violence and armed robbery pose serious threats to public security, including in Indonesia. The rise in crimes involving assault, mob violence, and the misuse of sharp weapons and firearms demands a more effective and responsive surveillance system. Manual CCTV-based surveillance has limitations, primarily due to officer fatigue and the potential for errors in detecting threats in real time. This study proposes the development of a CCTV-based surveillance system using the YOLOv11 algorithm that can detect acts of violence and the presence of weapons simultaneously in a single, integrated system. The system is developed as a desktop application and equipped with a Telegram notification mechanism as an early warning system to provide early warnings when valid threats are detected. Evaluation results show that the YOLOv11 model achieved good performance with an overall mean Average Precision (mAP) value of 0.914, and precision and recall values of 0.906 and 0.868, respectively. The system is capable of real-time detection with stable performance in the range of 30–40 FPS, supports live video processing and file uploads, and sends Telegram notifications in less than two seconds after a threat is detected. These results indicate that the developed system can function as a supporting system for video-based security surveillance.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Deteksi Senjata, Deteksi Kekerasan, CCTV, YOLOv11, Early Warning System, Weapon Detection, Violence Detection, CCTV, YOLOv11, Early Warning System
Subjects:	T Technology > T Technology (General) > T59.7 Human-machine systems.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 59101-(S2) Master Thesis
Depositing User:	Muhammad Syarif Januriansyah
Date Deposited:	29 Jan 2026 02:56
Last Modified:	29 Jan 2026 02:56
URI:	http://repository.its.ac.id/id/eprint/130977

Actions (login required)

View Item