Fruit Feature Recognition Based on Unsupervised Competitive Learning and Backpropagation Algorithms

This paper presents Fruit Feature Recognition (FFR) based on Unsupervised Competitive Learning (UCL) and Back-propagation Neural Network (BPNN) techniques. Fruit image data set comprising 300 samples categorized into three fruit classes namely Apple, Banana and Mango were used. Fruit attributes were extracted using Gray Level Co-occurrence Matrix (GLCM) functions. The applied simulations showed that UCL and BPNN algorithms were able to recognize the fruit images. The performance comparison tests further showed that the UCL was able to achieve a higher accuracy compared to the BPNN explorations which attained acceptable degree of recognition. The experimental simulation analysis of fruit recognition system achieved an accuracy of 94% for UCL and 90% for the BPNN.