POINT OF SALE SCANNING THROUGH YOLOV8 MULTI-OBJECT DETECTION IMPLEMENTATION FOR RETAIL PRODUCT RECOGNITION

Wicaksana, Alfa Aditya (2024) POINT OF SALE SCANNING THROUGH YOLOV8 MULTI-OBJECT DETECTION IMPLEMENTATION FOR RETAIL PRODUCT RECOGNITION. Other thesis, UPN Veteran Yogyakarta.

[thumbnail of Cover.pdf] Text
Cover.pdf

Download (1MB)
[thumbnail of Abstrak.pdf] Text
Abstrak.pdf

Download (142kB)
[thumbnail of Halaman Pengesahan.pdf] Text
Halaman Pengesahan.pdf

Download (973kB)
[thumbnail of Daftar Isi.pdf] Text
Daftar Isi.pdf

Download (192kB)
[thumbnail of Daftar Pustaka.pdf] Text
Daftar Pustaka.pdf

Download (220kB)
[thumbnail of Laporan Tugas Akhir.pdf] Text
Laporan Tugas Akhir.pdf
Restricted to Repository staff only

Download (11MB)

Abstract

Point-of-sale (POS) systems play a crucial role in modern retail by streamlining
transactions and improving operational efficiency. However, traditional barcode-based
scanning methods can become bottlenecks, particularly in high-traffic retail environments.
Small businesses, such as warungs in Indonesia, often lack access to automated checkout
systems, leading to inefficiencies in transaction processing. This study explores the
implementation of YOLOv8, a deep-learning-based object detection model, to enhance POS
scanning by enabling the simultaneous detection of multiple retail products in a single image.
In this study, a dataset of 35 Indonesian retail product categories was collected and
annotated for training. The YOLOv8m model was implemented and trained using a
supervised learning approach, employing data augmentation techniques to improve
generalization. The model was trained and validated using the Kaggle platform, with
evaluation metrics such as precision, recall, and mean average precision (mAP) used to
assess detection performance. The system was tested under various conditions to measure
accuracy, response time, and robustness to environmental factors.
The results demonstrate that YOLOv8 achieves high accuracy in retail product
recognition, with precision, recall, and mAP scores exceeding 90%. The model effectively
detects multiple objects within a single image while maintaining real-time processing
capabilities. Tests under different lighting conditions and product arrangements within
realistic limits indicate that detection performance remains stable, highlighting the model’s
robustness.
The findings of this study indicate that YOLOv8 is a viable solution for improving
POS scanning in small retail environments, such as warungs. The implementation of this
system has the potential to enhance efficiency and reduce human error in retail transactions.
Future research may focus on optimizing the model for real-time deployment, integrating it
with mobile POS applications, and expanding its capabilities to handle dynamic retail
environments.
Keywords: warung, YOLOv8, object detection, retail products, point-of-sale

Item Type: Thesis (Other)
Uncontrolled Keywords: warung, YOLOv8, object detection, retail products, point-of-sale
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Divisions: Faculty of Engineering, Science and Mathematics > School of Electronics and Computer Science
Depositing User: A.Md Eko Suprapti
Date Deposited: 30 Apr 2025 06:25
Last Modified: 30 Apr 2025 06:25
URI: http://eprints.upnyk.ac.id/id/eprint/42494

Actions (login required)

View Item View Item