Machine Learning Enabled System Architecture for Automatic Grading and Sorting of Walnut Fruits: A Review

India is the eighth largest walnut producer worldwide and exports walnuts to 70 countries worth 10 million USD (778 million INR) annually. The grading of walnut ripeness is the main concern of worry for customers and the fruit business. The efficacy and fast classification of a walnut’s ripeness stage are crucial in judging the quality. Walnut maturity classification mainly influences walnut fruit quality and later marketing. Our approach to the automatic sorting and grading of walnut ripeness focuses on the identification of marks that appear on the outer surface of walnut fruits. The image dataset for the identification of marks or patches is divided into four different classes unripe, mid-ripe, ripe, and overripe acquired through high-quality cameras used in the proposed architecture, performing feature engineering and classification using ANN, SVM, and CNN techniques. The Raspberry Pi function counts fruits in a fruit box, using a conveyor belt to detect marks and patches for grading and sorting of walnut fruit based on texture, size, and color features. A motor powers the belt, moving walnuts from the fruit placer to the mechanical part. This architecture automates the fruit grading process, significantly reducing labor costs and time compared to manual sorting. This improves productivity and allows for faster processing of large quantities of fruits and is capable of quality control, increased efficiency, reduced errors, and improved market access.