EDL4CC: Ensemble-based deep learning for crop diseases early detection
Abstract
In this study, we proposed ensemble based deep learning approached to efficiently classify crop diseases. The current paper was inspired to improve the limitation of paper entitle with "computer vision approach using GPU infrastructure to classify wheat rust diseases". Under the quoted study, We have employed different pre-trained models such as Inceptionv3, Resnet50, and VGG19 to build our model. From the experiment results, the VGG19 model outperform the other model to classify wheat rust diseases with promising accuracy. Data size and model generalization capability were the main limitations. In this work, we proposed an ensemble-based deep learning approaches to optimize the classification performance of base learners.The aim of this work is to handle the limitations of the previous work. To conduct the experiment, more than 23 thousand dataset acquired locally and kaggle repositories. From the experimental results, the ensemble-based learning model classified diseases types with 99.48 accuracy. To further improve the performance of the proposed model, We have considered scalability and drifting aspect to monitor model performance over time.