Darmawan, Ridwan (2024) SISTEM DETEKSI DAN PERHITUNGAN FREKUENSI PENGUNJUNG YANG KELUAR/MASUK PUSAT PERBELANJAAN MENGGUNAKAN YOLOv8n SECARA REAL-TIME. Other thesis, UPN Veteran Yogyakarta.
Text
ABSTRAK.pdf Download (201kB) |
|
Text
COVER.pdf Download (171kB) |
|
Text
DAFTAR ISI.pdf Download (43kB) |
|
Text
DAFTAR PUSTAKA.pdf Download (108kB) |
|
Text
PENGESAHAN PEMBIMBING.pdf Download (549kB) |
|
Text
PENGESAHAN PENGUJI.pdf Download (458kB) |
|
Text
SKRIPSI FULL.pdf Restricted to Repository staff only Download (4MB) |
Abstract
vii
ABSTRAK
YOLOv8n digunakan dalam penelitian ini untuk mendeteksi pengunjung yang keluar,
masuk, dan berada di dalam pusat perbelanjaan. Objek yang dideteksi berfokus pada
manusia, dengan menggunakan dataset MS COCO yang beranotasi untuk kelas person.
Model YOLOv8n dipilih karena kecepatan dan efisiensinya yang tinggi dalam melakukan
deteksi real-time, menjadikannya ideal untuk aplikasi yang memerlukan pemrosesan cepat
dan akurat di lingkungan publik seperti pusat perbelanjaan.
Metode deteksi yang digunakan dalam sistem ini melibatkan model YOLOv8n untuk
mendeteksi objek manusia, serta ByteTrack sebagai metode pelacakan objek (tracking).
ByteTrack dipilih karena kemampuannya dalam melacak objek yang bergerak,
memungkinkan sistem tidak hanya mendeteksi pengunjung tetapi juga melacak pergerakan
mereka secara berkelanjutan. Kombinasi dari YOLOv8n dan ByteTrack memungkinkan
deteksi dan perhitungan pengunjung secara lebih akurat dalam kondisi real-time.
Hasil pengujian model menunjukkan bahwa YOLOv8n menghasilkan presisi sebesar
86.6%, recall sebesar 77.6%, mAP50 sebesar 87.4%, dan mAP50-95 sebesar 63.4%. Sistem
ini diuji menggunakan video uji yang diambil dari kamera iPhone 13, dan hasilnya
menunjukkan akurasi deteksi real-time antara 91.67% hingga 95.83%. Tantangan yang
dihadapi termasuk kesalahan deteksi karena objek yang bertumpuk serta false positives dan
false negatives. Oleh karena itu, perbaikan lebih lanjut diperlukan, termasuk peningkatan
algoritma deteksi. Sistem ini diharapkan dapat membantu manajemen pusat perbelanjaan
dalam mengevaluasi daya tarik serta pengelolaan ruang berdasarkan jumlah pengunjung.
Kata Kunci: Deteksi Objek, YOLOv8n, Perhitungan Pengunjung, Real-time, ByteTrack
viii
ABSTRACT
YOLOv8n is used in this study to detect visitors entering, exiting, and present within a
shopping mall. The detected object is focused on humans, utilizing the MS COCO dataset
annotated for the person class. YOLOv8n was chosen for its speed and high efficiency in
real-time detection, making it ideal for applications requiring fast and accurate processing
in public environments such as shopping malls.
The detection method employed in this system involves the YOLOv8n model for
detecting human objects, along with ByteTrack as the object tracking method. ByteTrack
was selected for its ability to track moving objects, allowing the system not only to detect
visitors but also to continuously track their movements. The combination of YOLOv8n and
ByteTrack enables more accurate visitor detection and counting in real-time conditions.
The model testing results show that YOLOv8n achieved a precision of 86.6%, recall of
77.6%, mAP50 of 87.4%, and mAP50-95 of 63.4%. The system was tested using a video
captured by an iPhone 13, and the results demonstrated real-time detection accuracy
ranging from 91.67% to 95.83%. Challenges encountered included detection errors due to
overlapping objects, as well as false positives and false negatives. Therefore, further
improvements are necessary, including enhancing the detection algorithm. This system is
expected to assist shopping mall management in evaluating attractiveness and optimizing
space based on the number of visitors.
Keywords: Object Detection, YOLOv8n, Visitor Counting, Real-time, ByteTrack
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Object Detection, YOLOv8n, Visitor Counting, Real-time, ByteTrack |
Subjects: | Z Bibliography. Library Science. Information Resources > ZA Information resources |
Divisions: | Faculty of Engineering, Science and Mathematics > School of Engineering Sciences |
Depositing User: | Eko Yuli |
Date Deposited: | 01 Nov 2024 02:23 |
Last Modified: | 01 Nov 2024 02:23 |
URI: | http://eprints.upnyk.ac.id/id/eprint/41510 |
Actions (login required)
View Item |