Wheat Disease Detection And Growth Stage Monitoring Using Deep Learning Architectures

Traditional methods for diagnosing illnesses in wheat plants rely on visual inspections by experts, which is a tedious task. To enhance disease detection in plants, we propose a novel approach called Enhanced Vision CNN, which combines the multi-head attention feature of a vision transformer with the feature extraction powers of a CNN. Combining these methods allows the model to properly describe image features, and effectively capture both global and local information. Furthermore, to overcome limitations in conventional growth monitoring methods, we employ an ensemble approach using DenseNet201, InceptionV3, and InceptionResNetV2 models to track the growth stages of wheat plants. Our enhanced vision CNN model performed admirably in experimental findings, detecting diseases with a 99.4% accuracy. The ensemble models also obtain a growth phase detection accuracy of 88.5%. These results demonstrate the revolutionary potential of our methods for disease diagnosis and growth tracking in wheat plants.