Application of Deep RESNET Architectures for Plant Disease Classification Based on Visual Features

This article focuses on the application of the ResNet-50V2 neural network for automated plant disease detection. The modified ResNet-50V2 architecture incorporates enhanced residual connections, attention modules and a multitask loss function, improving diagnostic accuracy and reducing the risk of overfitting. The PlantVillage dataset containing over 16,000 images of both healthy and diseased plant leaves was utilized. Experimental results demonstrated that the ResNet-50V2-based model outperformed ResNet-50V1 in classifying plant pathologies. These findings were validated using precision, recall and F1-score metrics. The study concludes that the ResNet-50V2 can be effectively applied in agronomic tasks to reduce plant disease diagnosis time, offering valuable insights for agricultural automation specialists aiming to optimize costs and overcome productivity barriers in farming.