Leveraging Deep Learning-based EfficientNet Model for Cassava Leaf Disease Detection

Abstract

Agriculture is the primary means of subsistence for a section of India’s population. It is a vital component of enhancing the country's economy and growth. The demand for food has increased due to the exponential growth of the world's population. Progressively, farmers are adopting various farming techniques that use artificial intelligence to meet this need and increase production. Crop health observation, plant disease recognition and discovery, and weather and commodity price forecasts are some of the most common uses of artificial intelligence. Manually identifying crop diseases, particularly on bigger farms with the assistance of specialists, may be a tedious and costly ordeal, and they constitute a danger to food safety. To tackle this problem, deep-learning methods provide automated procedure oversight, inspection, and robot guidance based on images, which can effectively control pests and diseases. In the proposed work, a cassava leaf disease detection model based on the EfficientNet model has been developed. The cassava leaf disease categorization was performed using a Kaggle dataset. The accuracy of the proposed model was resulted as 92.83%, whereas, the lowest loss of 0.2019 at epoch 16 has been analyzed.