A multi-scale feature fusion neural network for multi-class disease classification on the maize leaf images.

Maize is a globally important cereal crop, however, maize leaf disease is one of the most common and devastating diseases that afflict it. Artificial intelligence methods face challenges in identifying and classifying maize leaf diseases due to variations in image quality, similarity among diseases, disease severity, limited dataset availability, and limited interpretability. To address these challenges, we propose a residual-based multi-scale network (MResNet) for classifying multi-type maize leaf diseases from maize images.