Convolution Neural Network (CNN) Based Computerized Classification of Adulterated Fruits with SIFT and Bag of words (BOW)

Most of the minerals and vitamins can be obtained by humans with the intake of fruits. Fruits are very rich in fibre and health-boosting oxidants can be provided by them. With a daily routine of eating fruits, a human can avoid the risk of diabetes, cancer, heart diseases. But these days fruits are being adultered for the [1] economic and technical benefits. Adulteration is a kind of illegal act of mixing some harmful chemicals into the fruits in order to increase profits. Eating adultered fruits can lead to chronic diseases diarrhoea, cancer, liver disorder, and heart diseases. Adulteration is a serious threat to the health of humans. Even though the adulteration of fruits is banned, many traders are applying adulteration in the thirst of more profits. It is very difficult to identify the fruits which undergone adulteration with naked eyes. Image processing techniques can be applied for detecting the adultered fruits. In this paper, A Convolution Neural Network (CNN) with K-means clustering and Bag of words based technique is proposed to detect the fruits which undergone adulteration. Accuracy of 0.94 is achieved with the proposed system.