A Deep Learning Approach to Bean Leaf Disease Classification: Evaluating ResNet152V2 Performance on Angular Leaf Spot and Bean Rust

Effective disease control and sustainable agriculture in the field of plant pathology depend on early and precise diagnosis of bean leaf diseases. This work uses the ResNet 152V2 deep convolutional neural network model to classify bean leaf damage. Source from Kaggle, the dataset consists of three classed images: angular leaf spot, bean rust, and healthy. This work uses ResNet152V2, a strong architecture well known for managing challenging image classification challenges, to reach great accuracy in spotting these disorders. Trained on 1,034 photos, the model is verified with 133 images producing an overall accuracy of 88% on the validation set. Among other performance criteria, precision, recall, and F1-score range from 0.85 to 0.90, so demonstrating the significant capacity of the model to separate the several leaf states. Minimal misclassification is indicated by the confusion matrix analysis; Effective generalizing to unseen data is supported by training, validation loss and accuracy analysis. The results highlight how well deep learning models might improve early disease diagnosis, lower pesticide use, and enhance sustainable agriculture methods. Expanding the dataset, evaluating other model architectures, and creating real-time diagnostic tools to help to further enhance disease categorization and management in agriculture will be subjects of future work.