Plant Disease Classification using CNN-LSTM Techniques

Crops have implications for global food security and the agrarian society. Their rapid recognition and prevention improve the probability of implementing effective interventions, and this is why our society is interested in finding automated systems that can do so. The CNN-LSTM classifier is proposed in this study to detect plant diseases. This study has deep feature extraction from several fully linked layers in transfer learning. The collected features are provided as input into the LSTM layer to build a hybrid model for plant disease identification. The LSTM layers’ output predictions chose the input images’ class labels. The trials are run with data from the PlantVillage dataset, and the accuracy is computed to assess performance. The experiment results reveal that the proposed method produces outcomes enhanced than pre-trained deep architectures with an accuracy of 99.17%.