Traditional Machine Learning-Based Classification of Cashew Kernels Using Colour Features

Cashew is one of the major commercial commodities contributing to national economy Tanzania as foreign revenue. And yet still processing cashew run locally using manual labour for a big part. If processed well under ideal conditions, cashews kernels are expected be white in colour. But due various factors like prolonged roasting steam chambers or over-drying, some tend have slight brown colour, and these referred scorched cashews. Despite sharing same characteristics with kernels, including nutritional quality, supposed graded differently. In many places around world, particularly Tanzania, sorting grading process performed by hand. international trade, very important this means more effective consistent methods need applied stage production order increase quality products. The objective study was evaluate use traditional Machine Learning techniques classification colour features. experiment, features were extracted from images. include (μ), standard deviations (σ), skewness (γ) channels RGB HSV spaces. relevant problem selected applying wrapper approach Boruta Library Python, irrelevant ones removed. 5 models studied their efficiencies analysed. Logistic Regression, Decision Tree, Random Forest, Support Vector K-Nearest Neighbour. Tree model recorded least accuracy 98.4%. maximum 99.8% obtained Forest 100 trees. Due simplicity application high accuracy, recommended best study.