Research on the Application of Quantum Convolutional Neural Network in the Rapid Classification and Detection of Food Appearance Defects

Abstract

In recent years, deep learning-based appearance detection technology has been widely applied in the field of appearance detection. Applying it to the field of food appearance detection is of great significance for achieving intelligent and flexible detection of food appearance defects. The quantum convolutional neural network representation layer, hidden layer neuron model, and hidden layer neuron model were designed. Through optimizing the quantum rotation angle and neural connection weights by training the error function, a food appearance defect classification system based on quantum convolutional neural network was built. The appearance of apples was used as the detection object, and image samples were collected through offline collection methods. The experimental results show that the model proposed in this paper achieved test accuracy rates of 97.16%, 91.85%, 94.23%, 97.72%, and 95.43% for five different appearance quality types of apples (good, scratches, fruit rust, insect damage, and rot), and the overall classification recognition accuracy reached 95.41%. It demonstrated excellent performance in the task of food appearance quality classification. In the process of food appearance defect detection, it solved the problems existing in manual inspection and met the intelligent requirements for defect classification in the inspection process.