Comparative Evaluation of Deep Learning Models for the Identification of Plant Diseases

Plant diseases are a major challenge for worldwide agriculture causing both financial losses and lowered yields in crops. Successful management and prevention of plant diseases require an early detection to be done. In this study, comparative research on CNNs (Convolutional Neural Networks) and Cnn + recurrent neural network (RNN) architecture-based plant disease detection has been carried out. Built upon CNN model which is good at capturing features in images, the CNN+RNN model in use combine both quality of CNNs after feature extraction as well as temporal data handling capabilities of RNNs. The model is trained on and evaluated with benchmark dataset of plant disease images. Finally, we evaluate their performance using performance metrics (accuracy, precision recall; F1-score etc.) The results have showcased the advantages and drawbacks of each model for the use case for real agriculture scenarios.