Retail Management on Mobile Application using Product Classification

A retail store management system is a retail store that uses stock management and orders through the basket by adding a function to classify products from images to facilitate customers and create competitive efficiency for traditional retailers. Product classification is used to improve the performance of traditional retailers by classifying products within a modern retail store by using deep learning on product images to create a model for mobile applications. The best model is selected by comparing the results from learning and evaluating models based on MobileNetV2, VGG16, CNN, InceptionV3 and ResNet. Then, mobile applications have developed by using the best model to support product management for modern retailers. In a mobile application for modern retail management, there are users consisting of retailers and customers, whose functions include stock arrangement, shopping cart arrangement, and product classification. A mobile application is developed by implementing Flutter with the Dart language. It is a cross-platform mobile application development tool that uses Firestore as a database on the Firebase cloud.The results of the five models, it was found that the MobileNet, VGG16, CNN, InceptionV3 and ResNet models validity were 92.80, 80.79, 55.67, 49.68, and 04.59 respectively. The model with the highest accuracy was developed as a system for modern retail stores. The system has been tested for 6 months. The retailers 1, 2 and 3 have the highest number of products sold in the 1st, 6th, and 4th months, respectively.