Automated Classification of Fruits: Pawpaw Fruit as a Case Study

Abstract

Several fruit recognition researches have been performed on so many fruits cutting across orange, banana, strawberry, apple, etc. But little or no consideration has been given to pawpaw. This might be due to its complex nature. Identification of ripe pawpaw from unripe species using colour-based feature extraction method and back propagation neural network (BPNN) model is a complex task since a green looking pawpaw could be ripped when analysed. In this paper, the ripeness of pawpaw was determined by a simple colour recognition algorithm using a BPNN model. RGB colour components of the captured images are extracted after the pawpaw images are resized. After the application of a simple heuristic method, the colour components of the resized images are rescaled. Finally, the rescaled images’ colour histogram is obtained and used as feature vector and the BPNN model uses the feature vector to classify the pawpaw species. The proposed model has an accuracy of 97\(\%\).