Deep Classification: Ensemble Learning Models for Multi-Plant Disease Detection

Plant diseases play a significant role in India by improving agriculture production and food security. Early detection and prediction of plant diseases prevent heavy economic losses and increase agriculture. This study uses an ensemble learning model to classify multiple-plant diseases and a deep learning (DL) technique to increase the classification. A proposed framework in combination with transfer learning using pre-trained Inception in ensemble architecture of feature extraction capabilities. In this work, advanced preprocessing techniques such as AHE and guided image filtering are used for noise filtering. The segmentation method is like watershed segmentation. Finally, the proposed solution grouping combines Deep Belief Networks (DBNs) and softmax layers (for classification). The recall suggests the performance of the proposed approach on three datasets for plant diseases. Evaluation results show that the proposed classification rate and the accurate abnormal region detection rate are more than 99.78%.