Classification of Vegetable Quality with InceptionV3, Xception and Combination of These Models

With increasing emphasis on nutrition, food safety and consumer preferences, accurate assessment of the quality of vegetables has become important. Quality classification involves evaluating various characteristics such as appearance, texture, taste, nutritional content and shelf life. This process not only affects consumer satisfaction, but also the marketability, profitability and overall sustainability of the agricultural sector. With the use of deep learning methods in solving various problems today, their use has become widespread in the field of plant production. In this study, VegNet vegetable image dataset was used with InceptionV3, Xception and a model consisting of the combination of two models. According to the results of the study, the highest accuracy rate was obtained in the proposed model with ${9 6. 9 1 \%}$. Then, InceptionV3 with 95.01% accuracy rate and Xception model with 94.8% accuracy rate were determined. As a result, it is suggested in this study different models can be used together experimentally to improve the performance of classification models. In the study, time comparison was also made. Time comparison results show that the InceptionV3 model is better than the other models in terms of time. In terms of epoch time, proposed model has approximately the total epoch time of the two models.