Automatic People Detection and Tracking System Melalui Cctv Menggunakan Face Recognition pada Departemen Teknologi Informasi

Jauhar, Dava Aditya (2024) Automatic People Detection and Tracking System Melalui Cctv Menggunakan Face Recognition pada Departemen Teknologi Informasi. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05311940000030 - Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2026. Download (3MB) | Request a copy

Abstract

Penggunaan kamera pengawas CCTV merupakan salah satu metode pengamanan yang sering ditemui di berbagai lokasi. Penggunaaan kamera CCTV dapat membantu pengawasan dan pengumpulan bukti apabila terjadi kejadian kejahatan maupun kejadian lain yang tidak diinginkan. Meskipun demikian, seringkali proses pengumpulan dan pengolahan informasi dari kamera CCTV masih dilakukan menggunakan metode konvensional, yaitu melihat ke dalam rekaman kejadian yang telah terjadi dan mencari jejak dari kejadian tersebut yang tentu menghabiskan waktu dan tenaga dalam pencariannya. Selain itu, ketepatan dari proses ini bergantung kepada manusia yang mencari informasi. Pada tugas akhir ini, diajukan sebuah sistem pengenal, pencatat, dan pelacakan seseorang pada rekaman CCTV menggunakan teknologi face detection dan recognition untuk mengenali wajah seseorang dan melakukan pencatatan informasi seperti: identitas, lokasi, dan waktu kemunculan. Adapun pada sistem ini tidak diperlukan pengumpulan dataset wajah terlebih dahulu karena sistem akan membuat dataset secara real-time seiring dengan berjalannya sistem pengenalan wajah. Sistem yang dikembangkan menggunakan YOLOv8 sebagai model object detection, ByteTrack dan BOT-SORT sebagai model object tracking, dan ArcFace sebagai model face recognition. Hasil pengujian menunjukkan waktu inferensi tersingkat dan FPS tertinggi didapatkan oleh sistem dengan model Buffalo_sc untuk face recognition dan YOLOv8s dengan ByteTrack untuk object detection dan tracking. Dalam pengujian pengenalan wajah, sistem berhasil mendapatkan nilai accuracy sebesar 0,5, precission sebesar 0,5, dan recall sebesar 0,2. Meskipun demikian, masih terdapat beberapa kesalahan deteksi baik dari model face detection maupun face recognition.
=============================================================================================================================
The use of CCTV surveillance cameras is one of the security methods that are often found in various locations. The use of CCTV cameras can help supervise and collect evidence in the event of a crime or other unwanted events. However, often the process of collecting and processing information from CCTV cameras still uses conventional methods, i.e. looking into recordings of events that have occurred and looking for traces of the incident, which is certainly time-consuming and labor-intensive. of the incident which certainly consumes time and energy in the search. searching. In addition, the accuracy of this process depends on the human who is searching for information. who are looking for information. In this research, a system for recognizing, recording, and tracking of a person on CCTV footage using face detection and recognition technology to recognize a person's face and record information such as: identity, location, and time of appearance is proposed. As for this system, there is no need to collect face datasets in advance because the system will create datasets in real-time as the face recognition system runs. The developed system uses YOLOv8 as an object detection model, ByteTrack and BOT-SORT as object tracking models, and ArcFace as a face recognition model. The test results show that the shortest inference time and highest FPS are obtained by the system with the Buffalo_sc model for face recognition and YOLOv8s with ByteTrack for object detection and tracking. In face recognition test, system succeeded in getting an accuracy value of 0.5, precission of 0.5, and recall of 0.2. However, there are still some detection errors from both the face detection and face recognition models.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	ArcFace, CCTV, Face Recognition, Object Detection, Object Tracking, YOLO
Subjects:	T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Dava Aditya Jauhar
Date Deposited:	16 Feb 2024 08:19
Last Modified:	16 Feb 2024 08:19
URI:	http://repository.its.ac.id/id/eprint/107220

Actions (login required)

View Item