Cassava leaf disease classification using Deep Learning

Main Article Content

S. Aravind, S. Harini, Varun kumar K A

Abstract

Cassava is an important crop that is cultivated in the tropics. The Food and Agriculture Organization of the United Nations (FAO) states that Cassava is a source of food and income for more than 800 million people around the globe. And it is reported that Africa accounts for more than 50% of the global production among the other cassava-producing regions of the world. But several pests and infections have been a menace to the production of this crop for a long while now since they periodically ruin the crop in some regions leading to famine and consequently leaving the farmers in peril. This dire situation entails new methods to identify cassava diseases at an early stage and thus help prevent this crisis. In this Research, a deep CNN model implementing EfficientNet-B3 has been developed that identifies five main classes: Cassava Mosaic Disease (CMD), Cassava Bacterial Blight (CBB), Cassava Green Mottle (CGM)), Cassava Brown Streak Disease (CBSD and a healthy specimen. This model makes use of the compound coefficient, a different scaling approach prescribed for EfficientNet, to scale up the model in a more structured way. This approach thus avoids brute force manual tuning and stagnating performance issues. The model achieved an overall accuracy of 96.74%.

Article Details

Section
Articles