Fruits classification by using machine learning - An experiment using popular approaches on local data

Artificial intelligence is becoming more and more popular. It can be applied in so many different areas of our lives. Image processing is used for many different purposes such as traffic monitoring, security analysis, education, etc. In this paper, we study the image classification techniques to apply in the classification of fruits. The results of this study can be used to place fruit on corresponding shelves in supermarkets, detect mismatches of fruit on supermarket shelves, or check fruit prices without barcode scanning. In this study, we use three popular classification models including Random Forest and K-Nearest Neighbors (KNN), Support Vector Machine (SVM). Three feature types are used, including Histogram of Oriented Gradients (HOG), RGB histogram (RGB) and Canny edge detection (Edge). The evaluation results show that SVM model with HOG has the best accuracy with 96%.