Deep Learning Based Crop Yield Prediction and Disease Identification

The research focuses on two important areas of agriculture: crop yield forecasts and the categorization of plant diseases. We suggest using Graph Convolutional Networks (GCN), a deep learning method that makes use of the spatial correlations in plant photos to precisely identify illnesses, for the categorization of diseases. Through the analysis of these linkages, the GCN may improve agricultural production and sustainability by facilitating the early diagnosis and prevention of crop damage. We simultaneously create a time series model for agricultural yield forecasting that is based on Transformers. This model includes historical yield values and pertinent contextual information, and it displays historical crop yield data as time series sequences. Our goal is to achieve high accuracy crop production predictions in the future by training the model using historical data. The trained model shows that this strategy is feasible, achieving 90.35% accuracy on a held-out test set. All things considered, the method of using publicly accessible, progressively larger picture datasets to train deep learning models offers a direct route to widespread, smartphone-assisted crop disease detection.