Open Access

ARTICLE

Design Features of Grocery Product Recognition Using Deep Learning

E. Gothai^1,*, Surbhi Bhatia², Aliaa M. Alabdali³, Dilip Kumar Sharma⁴, Bhavana Raj Kondamudi⁵, Pankaj Dadheech⁶

1 Department of Computer Science and Engineering, Kongu Engineering College, Perundurai, Erode, 638060, Tamil Nadu, India
2 Department of Information Systems, College of Computer Science and Information Technology, King Faisal University, Riyadh, 11533, Saudi Arabia
3 Faculty of Computing & Information Technology, King Abdulaziz University, Rabigh, 21911, Saudi Arabia
4 Department of Mathematics, Jaypee University of Engineering and Technology, Guna, 473226, Madhya Pradesh, India
5 Department of Management Studies, Institute of Public Enterprise, Hyderabad, 500101, Telangana, India
6 Department of Computer Science and Engineering, Swami Keshvanand Institute of Technology, Management & Gramothan, Jaipur, 302017, Rajasthan, India

* Corresponding Author: E. Gothai. Email:

Intelligent Automation & Soft Computing 2022, 34(2), 1231-1246. https://doi.org/10.32604/iasc.2022.026264

Received 20 December 2021; Accepted 25 January 2022; Issue published 03 May 2022

Abstract

At a grocery store, product supply management is critical to its employee's ability to operate productively. To find the right time for updating the item in terms of design/replenishment, real-time data on item availability are required. As a result, the item is consistently accessible on the rack when the client requires it. This study focuses on product display management at a grocery store to determine a particular product and its quantity on the shelves. Deep Learning (DL) is used to determine and identify every item and the store's supervisor compares all identified items with a preconfigured item planning that was done by him earlier. The approach is made in II-phases. Product detection, followed by product recognition. For product detection, we have used You Only Look Once Version 5 (YOLOV5), and for product recognition, we have used both the shape and size features along with the color feature to reduce the false product detection. Experimental results were carried out using the SKU-110 K data set. The analyses show that the proposed approach has improved accuracy, precision, and recall. For product recognition, the inclusion of color feature enables the reduction of error date. It is helpful to distinguish between identical logo which has different colors. We can achieve the accuracy percentage for feature level as 75 and score level as 81.

---

Keywords

---