Detection of olive color and texture tissue with artificial intelligence models

In recent years, there has been a rapid increase in interest towards image processing and automation systems in the agricultural sector to improve productivity and quality standards. In traditional olive sorting methods, the process of color and quality detection performed by the human eye is time-consuming and prone to errors. In this thesis, an image processing system has been developed to sort raw olives in real-time based on their color (red, green, black) and quality (fly damage) status. The developed system aims to reduce labor costs and consistently maintain quality by performing accurate color and texture analysis of the olives, providing high accuracy and fast classification . Furthermore, this system integrated with industrial automation provides increased efficiency in large-scale production and minimizes the dependency on human labor in olive farming. The study not only focuses on olive sorting but also offers an applicable infrastructure for the quality control processes of other agricultural products, emphasizing the importance of modern technologies in the agricultural sector. The CNN architecture achieved an accuracy of 99.56% and demonstrated the highest performance with F1 scores of 0.9973, 0.9973, and 0.9935 for the Fly, Green, and Black classes, respectively. These results indicate that the model performs in a balanced and reliable manner across all classes.