Adaptive Feature Learning for Robust Pathogen Detection in Plants

This research presents an innovative deep learning method for classifying plant pathogens by using domain-adaptive feature extraction. The ResNet50 model has been adjusted and trained on the Plant Village dataset specifically for the purpose of extracting features. The retrieved characteristics are then inputted into a concise Deep Neural Network (DNN) with two hidden layers. The integrated model is then trained on a distinct dataset that is especially targeted against plant pathogens. The trial outcomes show how successful the suggested strategy is. Combined model achieves competitive results compared to a directly fine-tuned ResNet50 model on the plant pathogen dataset, highlighting the potential of domain adaptation for improving plant pathogen classification, particularly in scenarios with limited labeled data.