Abstract—Goods management in supermarkets is a simple yet time and effort consuming task. To mitigate the problem, we propose a computer vision based goods management system for supermarkets using Tiny YOLOv3 model, Convolutional Neural Network and Keras. The system, with a moving camera, can detect supermarket products such as cans, bottles, bags, etc. as well as their corresponding prices and barcodes thus allowing inventory management, empty shelves and mispricing detection, etc. The system was trained on our own dataset (object detection, 16 classes, using transfer training) and the SVHN dataset (digit recognition). Experiment results show that the system works well with 92.81% mAP (mean Average Precision) for object detection and 92.28% of accuracy for digit recognition.
 

Index Terms—object detection, digit recognition, Tiny YOLOv3, CNN, goods management

Cite: Huu-Thieu Do and Viet-Cuong Pham, "Deep Learning Based Goods Management in Supermarkets," Journal of Advances in Information Technology, Vol. 12, No. 2, pp. 164-168, May 2021. doi: 10.12720/jait.12.2.164-168

Copyright © 2021 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.