Lemon (Citrus Limon) Quality Classification with Convolutional Neural Network (CNN) and Transfer Learning VGG16

Lemon fruit is a fruit that is rich in vitamin C and can be found in various countries. However, lately, lemons have decreased due to climate change and high production costs. One of the causes of high production costs is sorting good quality lemons and bad quality lemons, which is done manually, that is, by working with many people and taking a long time so that the production costs of lemons are prohibitive. Automatic sorting using intelligence is considered an effective artificial way to speed up the sorting of good quality lemons and bad quality lemons so that it can be done quickly and can reduce the production costs of lemons. Then, to make an artificial intelligence model with good accuracy, it is necessary to consider several factors, namely the development method, data separation, and the algorithm used. In this study, the development method used was the Cross-Industry Standard Process for Data Mining (CRISP-DM) with the Convolutional Neural Network (CNN) algorithm and VGG16 Transfer Learning with the scenarios of $90: 10,80: 20$, and 70:30 data separation from the results of making the model for the highest accuracy in the $90: 10$ data splitting scenario with the CNN Algorithm. Using the confusion matrix, an accuracy value of $\mathbf{9 9. 6 3 \%}$, precision of $\mathbf{9 9. 5 1 \%}$, recall of $\mathbf{9 9. 8 0 \%}$, and $\mathbf{F - 1}$ Score was obtained with a value of $\mathbf{9 9. 6 6 \%}$. Then, the CNN model was successfully integrated with mobile-based applications.