Weight Prediction of Durum Wheat Grains Using Computer Vision and Machine Learning

Abstract

Weight of the grains is an important feature in the quality assessment and agricultural mechanization processes. This study presents a method for predicting the weight of durum wheat grains by combining image processing techniques with artificial neural networks (ANN). Physical characteristics of wheat grains, such as length, width, thickness, and projection area, were extracted from high-resolution images and used as inputs for the ANN model. The model’s performance was evaluated using several statistical metrics, including Mean Squared Error (MSE), Root Mean Squared Error (RMSE), and R-squared (R²). The ANN achieved an average MSE of 1.216×10- ⁵ and RMSE of 3.34×10- ³ grams, with an R² value of 0.89, indicating high predictive accuracy. These results demonstrate that the integration of image processing and ANN can effectively estimate wheat grain weight with minimal error, offering a practical solution for applications in agricultural mechanization.