A Hybrid Neural Network Based Maize Leaf Disease Identification Integrating ResNet50 and Attention Mechanism

Abstract

Even though maize is an important global staple diet, viral diseases on leaves threaten its productivity leading to significant yield loss. Correct and timely detection of these diseases is crucial in effective crop management. This paper discussed, an advanced deep neural network (DNN) for accurate recognition of maize leaf diseases (Gray leaf spot, Common rust, and Tar Spot). Using ResNet50, a potent feature extractor coupled with an attention mechanism results in increased focus on the areas disease specific. This fusion seeks to increase the overall accuracy of identification. ResNet50 gets complex features out of images, allowing it to recognize complicated disease characteristics. The attention mechanism allows the model to focus on crucial image areas, which makes it possible for raising interpretability and robustness. The experiments validation over a generalized data set would verify the model’s better efficiency and confirm its role as an accurate tool for precision agriculture. So, to conclude its improvements detect crop diseases and provide a reliable tool for the precision diagnosis of maize leaves through combining ResNet50 with an attention model.