Application Research of Non-destructive Detection of Apple Sugar Content Based on Convolution Neural Network

This paper studies the application of non-destructive testing of apple sugar content based on convolutional neural network (CNN). The Apple sample dataset was constructed and a specific network structure was designed based on VGG16, the well-trained network model was transplanted to the Android mobile phone, and the feature information related to Brix in the Apple image was successfully extracted by taking photos on the mobile phone. Experiments show that the Brix prediction method based on convolutional neural network performs well in indicators such as accuracy, recall and F1 score. Its application potential in fruit quality inspection is verified.