AUTOMATED CLASSIFICATION OF FRUIT IMAGES

Abstract

Main objective of this project is to build image recognition algorithms to identify fruits from image feeding. This report explores different techniques used for the recognition of fruits. Dissimilar stages linking image processing like the pre processing stage, segmentation and feature extraction will be studied and discussed. At last the end product ofthe algorithms will be written in Python language. This project uses the Artificial Neural Network method to build the software. The key benefit of using this technique is that it offers features extraction and detection that is appropriate for character recognition. Dissimilar models of neural network are discussed and convolutional Neural Network with mobile net architecture was used. After numerous tests a suitable set oftraining parameters are describe and network structure that consist of(one input layer), (one hidden layer ) and (one output layer) with 230 input neurons, 115 neurons for the hidden layer and 108 neurons for output layer is formed. The system leading continues with the pre-process of the captured picture with picture flattening, colour correction and then feature extraction through image array. Following, the convolutional neural process over the network is raised to return an output matrix. Based on the output matrix, the recognized character can be determined. We worked on dissimilar data set and train them to capture and specify the image ofdifferent fruits, it give us 100% accuracy result through CNN algorithm.