A Machine-Learning Approach for Automatic Grape-Bunch Detection Based on Opponent Colors

This paper presents a novel and automatic artificial-intelligence (AI) method for grape-bunch detection from RGB images. It mainly consists of cascade support vector machine (SVM)-based classifiers that rely on visual contrast-based features that, in turn, are defined according to grape bunch color perception. Due some principles opponent theory proper contrast measures, precise estimate bunches is achieved. Extensive experimental results show the proposed able accurately segment grapes even uncontrolled acquisition conditions with limited computational load. Finally, such an approach requires very small number training samples, making it appropriate onsite real-time applications implementable smart devices, usable set up by winemakers.