Automatic Fruit Recognition from Natural Images using Color and Texture Features

Automated or robot-assisted harvesting is an emerging domain of research that combines the aspects of computer vision and machine intelligence. This research is usable in monitoring, sorting and picking of fruits for ensuring faster production chain. This paper aims to analyze popular methods of auto-harvesting, categorization of fruits and proposes a new approach that overcomes some of the drawbacks of the previous methods. The proposed approach takes into account different types of fruits. The main goal is to come up with a method for classifying these different types of fruits accurately and efficiently. Images are preprocessed in order to separate the fruit in the foreground from the background. Texture features from Gray-level Co-occurrence Matrix (GLCM) and statistical color features are extracted from the segmented image. Two types of features are combined in a single feature descriptor. A Support Vector Machine (SVM) classification model is trained using these feature descriptors extracted from the training dataset. Once trained, the model can be used to predict the category for an unlabeled image from the validation set. The proposed approach also works best for embedded systems and single board computers as it realizes the trade-offs of these devices.