Apple Quality Identification and Classification by Computer Vision Based on Deep Learning
Abstract This work researched apple quality identification and classification from real images containing complicated disturbance information (background was similar to the surface of the apples). This paper proposed a novel model based on Convolutional Neural Networks (CNN) which aimed at accurate and fast grading of apple quality. The proposed model was trained and validated, with best training and validation accuracy of 99% and 98.98% at 2590th and 3000th step, respectively. Two other methods, which were Google Inception v3 model and traditional imaging process method, were also used for apple quality classification. The greatest training accuracy of the Google Inception v3 model was 92% with 91.2% validation accuracy. The 78.14% accuracy was obtained by traditional method based on histogram of oriented gradient (HOG) and gray level co-occurrence matrix (GLCM) features merging and support vector machine (SVM) classifier. The three models were tested using independent 300 apples testing set, getting accuracy of 95.33%, 91.33%, and 77.67%, respectively. The results showed that the proposed model was more helpful and accurate for classification of apple quality. Furthermore, the training times of three methods were 27, 51, and 287 minutes, respectively. The proposed model can be considered a cost-effective method for fast grading of apple quality.