A Hybrid Deep Learning Approach for Severity Classification of Pepper Leaf Mosaic Virus

The efficient administration of agricultural operations and the maximization of crop output both depend critically on an accurate assessment of the severity of plant diseases. This study aims to answer this need by providing a hybrid deep-learning model for the categorization of the severity of the Pepper Leaf Mosaic Virus (PLMV). This model mixes Convolutional Neural Networks (CNNs) and Long Short-Term Memory networks (LSTMs). The suggested model makes use of the advantages offered by both the spatial feature extraction and the temporal dynamics modeling approaches. As a result, it is possible to gain an in-depth comprehension of the progression of PLMV severity. The study makes use of a dataset that contains twenty thousand photographs of pepper plants taken at various phases of PLMV infection. In order to guarantee precise classification, the architecture of the hybrid CNN-LSTM model has been painstakingly crafted to take into account the subtleties of the course of PLMV severity. The performance of the model can be made more reliable by doing exhaustive data preparation, which may include image scaling and enhancement. In order to prevent the model from becoming overfit, it is trained using stochastic gradient descent, which also includes dynamic learning rate scheduling and early stopping. The findings of this study shed light on the substantial potential that the hybrid CNN-LSTM model possesses in terms of the classification of PLMV severity. The findings may have repercussions for precision agriculture, which would benefit from having access to a more developed tool for making well-informed decisions and implementing timely interventions. The research lays the groundwork for further developments in illness severity assessment and management, despite the fact that it has some limitations, such as the fact that the accuracy of the model is affected by the influence of image quality and environmental conditions.