Neural Network Approach for Early Detection of Sugarcane Diseases

Almost 37% of land is used for agriculture purpose across the world and in fact 60% of arable land in India is used for agriculture purpose. Hence, agriculture plays an important role in the Indian economy growth and its present Gross contributed Domestic Products-GDP share for the country is 20.5% alone. The major4 cash crops grown in India are Sugarcane, cotton, jute, oil seeds and tobacco; this paper mainly concentrate on sugarcane crop as India is the second largest producer of sugarcane crop and also, it plays a prevalent role not only in modern diet but also pushed by the Indian government to extract ethanol to blend with petrol to generate bio-fuel so as to reduce the import of fuel from outside and also in the manufacture of bio-plastics. Also, Sugar industry is one of the significant agricultural sectors that impacts rural livelihood of about 50 million sugarcane farmers and around 5 lakh workers directly employed in sugar mills. This paper mainly discusses the variety of cane seeds available for sowing; various cane diseases and early detection of diseases using CNN based neural network methodology and its comparison with other techniques such as KNN, SVM, and RNN. The artificial neural network (ANN) algorithms are tested for mainly nine different classes including healthy crop and these approaches are verified in terms quality metrics like accuracy, F1 score, recall and precision and comparison of both the approaches are discussed.