Perancangan Awal Sistem Automatic Self-Checkout Untuk Produk Buah Berbasis CNN dan Sensor Berat Loadcell

Abstract


 
 
 The check-out system is one of the most important systems when we make purchases at a shopping center, shops, supermarkets, or mini markets. Determining different types of fruit is one of the problems faced in making payments, especially in supermarkets, because cashiers must determine the category labeling and weight of a particular fruit to determine the price. In this research, a design is proposed which utilizes Deep Learning in the field of image processing using the Convolutional Neural Network (CNN) algorithm utilizing RaspberryPi3. In making a classification system that uses deep learning, there are several main process stages, such as dataset collection, system design, training, and testing to see the results of the designed CNN model. The processed dataset in this research is a fruit image dataset derived from the Fruits-360. The results of the learning process obtained CNN models with 100% accuracy and loss of ≤ 0.026. The results of the CNN model that has been carried out will be combined with the load cell sensor which is expected to be used as an automatic self-checkout, especially for fruits. On the other hand, the calibration process of the loadcell weight measurement sensor is conducted which resulting an error measurement rate of 0.2 grams.