Afif, Muhammad (2026) Rancang Bangun Platform Smart Surveillance Dengan Integrasi Computer Vision Pada Edge Device Untuk Meningkatkan Efisiensi Pemantauan Tindak Kekerasan Fisik Di Lingkungan Kampus. Diploma thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5027221032-Undergraduate_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5027221032-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (9MB) | Request a copy |
Abstract
Tindak kekerasan di lingkungan kampus masih menjadi permasalahan yang serius di Indonesia. Data Kementerian Pemberdayaan Perempuan dan Perlindungan Anak menunjukkan terdapat 2.378 kasus kekerasan pada tahun 2024 dengan korban berjenjang pendidikan perguruan tinggi. Sistem CCTV konvensional yang ada saat ini mampu memperluas jangkauan pengawasan, namun masih memiliki keterbatasan dalam melakukan deteksi otomatis dan memerlukan pengawasan intensif dari operator manusia. Penelitian ini mengembangkan platform smart surveillance berbasis website yang mengintegrasikan computer vision pada edge device untuk meningkatkan efektivitas pemantauan tindak kekerasan fisik secara real-time di lingkungan kampus. Sistem dirancang untuk mendeteksi empat jenis kekerasan fisik yaitu assault, fighting, shooting, dan robbery menggunakan arsitektur model Spatial-Temporal Graph Convolutional Network yang dikombinasikan dengan Temporal Convolutional Network (ST-GCN-TCN) untuk klasifikasi kekerasan berbasis sekuens pose manusia, serta model YOLOv8n-Pose untuk ekstraksi skeleton. Model ST-GCN-TCN yang telah melalui proses hyperparameter tuning sebanyak 100 trials menggunakan Optuna menghasilkan nilai f1-score sebesar 0,9503 sebelum kuantisasi dan 0,9562 setelah kuantisasi INT8, namun akurasi menurun menjadi 0,30 ketika diterapkan pada lingkungan real-life. Sistem dilengkapi dengan fitur auto-reporting melalui WhatsApp dan email yang mampu mengirimkan laporan deteksi dalam waktu di bawah 20 detik. Hasil penelitian menunjukkan bahwa integrasi edge computing, computer vision, dan platform website mampu meningkatkan efisiensi pengawasan kampus secara signifikan dan memberikan peringatan dini terhadap tindak kekerasan di lingkungan kampus.
====================================================================================================================================
Violence in campus environments remains a serious concern in Indonesia. Data from the Ministry of Women's Empowerment and Child Protection recorded 2,378 cases of violence in 2024 involving victims at higher education levels. While conventional CCTV systems help expand surveillance coverage, they are limited in automatic detection capability and require intensive monitoring by human operators. This research develops a website-based smart surveillance platform that integrates computer vision on edge devices to improve the effectiveness of real-time physical violence monitoring in campus environments. The system is designed to detect four types of physical violence, namely assault, fighting, shooting, and robbery, using a Spatial-Temporal Graph Convolutional Network combined with a Temporal Convolutional Network (ST-GCN-TCN) architecture for skeleton-based violence classification, alongside a YOLOv8n-Pose model for skeleton extraction. The ST-GCN-TCN model, optimized through 100 trials of hyperparameter tuning with Optuna, achieved an f1-score of 0.9560 before quantization and 0.96 after INT8 quantization, but the accuracy decreases when implemented in the real-life environment. The system is equipped with an auto-reporting feature via WhatsApp and email capable of delivering detection reports to security personnel in under 20 seconds. The results demonstrate that the integration of edge computing, computer vision, and a unified website platform can significantly enhance campus surveillance efficiency and provide early warnings against acts of violence in the campus environment.
| Item Type: | Thesis (Diploma) |
|---|---|
| Uncontrolled Keywords: | Computer Vision, Edge Computing, Real-Time Monitoring, ST-GCN-TCN, Violence Detection, Computer Vision, Edge Computing, Real-Time Monitoring, ST-GCN-TCN, Violence Detection |
| Subjects: | T Technology > T Technology (General) |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis |
| Depositing User: | Muhammad Afif |
| Date Deposited: | 31 Jul 2026 04:27 |
| Last Modified: | 31 Jul 2026 04:27 |
| URI: | http://repository.its.ac.id/id/eprint/140644 |
Actions (login required)
 |
View Item |