Apple Grade Classification Based on Convolutional Neural Network and Machine Learning

The classification of apple grades is crucial for apple sales. Using computer vision technology as a non-destructive method to classify fruits has the advantages of high speed, applicability and high precision. This study focuses on the classification of apple grades using convolutional neural networks (CNN) and machine learning (ML) models. Firstly, the transfer learning method is adopted, and six popular CNN architectures (VGG-16, VGG-19, ResNet-50, MobileNet, EfficientNet, and Inception_V3) are used to compare the effects of different convolutional neural network models in processing apple classification. The results show that EfficientNet has the highest classification accuracy. Secondly, combining CNN and ML model, using EfficientNet to extract depth features and training ML models: K nearest neighbor (KNN), random forest classifier (RFC) and support vector machine (SVM) respectively. The results show that the classification accuracy is improved. Finally, the ensemble learning method is used to integrate multiple CNN_ML models through voting mechanism, and the final classification model is obtained, with an accuracy rate of 98.21%. According to the experimental results, this study was successful in apple grade classification and provided new ideas for apple grade classification.