Building an automatic agricultural products recognition system to facilitate the export process of fruits and vegetables

Agriculture plays a significant role in the Moroccan economy, providing ample employment opportunities and serving as a crucial export industry. In order to further enhance this sector, Morocco introduced the “Generation Green 2020-2030” plan in February 2020, which focuses on accelerating the digital transformation of agriculture. Hence, the aim of our research is to build an automatic system for recognizing fruits and vegetables to simplify the export process of agricultural products. The best state-of-the-art systems use Convolutional Neural Networks Or approaches to recognize agriculture products. In this paper, we propose a novel recognition scheme. To achieve optimal results, we conducted several experimental tests using various combinations of different machine learning literature approaches. The results show that the pre-process of data and the use of features such as the shape, the color, and the texture of the agricultural products, the projection of that features in a novel discriminant space using the Principal Component Analysis technique, and a post-classification using the Support Vector Machine approach lead to achieve a good accuracy performance