Fruit Recognition Based On Convolution Neural Network

Computer vision is widely used at present. However, fruit recognition is still a problem for the stacked fruits on weighing scale because of complexity and overlap. In this paper, a fruit recognition algorithm based on convolution neural network(CNN) is proposed. At first the image regions are extracted using selective search algorithm, then the regions have been selected by means of an entropy of fruit images, and finally these regions are regarded as input of CNN neural network for training and recognition. The final decision is made based on a fusion of all region classifications using voting mechanism. In order to achieve the actual application in supermarket, we have considered the variety of fruit, stack of fruits, the changes of fruit number and position, and have made a multifarious training set of fruits. After the network has been trained with an optimal training set, it has obtained a remarkable recognition rates for the fruits stacked on a weighing scale.