Mango Anthracnose and Powdery Mildew Disease Detection Using Convolutional Neural Network and Artificial Neural Network

Abstract

Mango is the third most important tropical fruit crop after banana and citrus. Konkan region is a large mango producing belt on the west coast of Maharashtra, accounts for around 10% of the entire land under mango in the country. Anthracnose and powdery mildew, caused by Colletotrichum gloeosporioides and Oidium mangiferae, respectively, are two important diseases impeding mango export and inflicting drastic yield losses in India. During the heavy wet season, losses from this disease were estimated to be 60% or greater. The diseasesdirectly reduce the amount and quality of collected products. Similarly, Powdery mildew is a widespread predisposing disease of panicles, bloom clusters, fruits, and foliage. Because of its effect on fruit set and development, the disease can reduce output by up to 70%. With the goal of managing plant disease with few inputs at an early stage, the current study intends to build disease detection models for anthracnose and powdery mildew of mango using Convolution neural networks (CNN) and Artificial neural networks (ANN). The teachable machine models developed using RGB images for detection of anthracnose and powdery mildew of mango and ANN model developed using thermal data of mango leaves infected with anthracnose disease performed very goodascomparedto the existing plant disease detectionmodels.