Ensemble Average and Deep Neural Networks for Detection of Rice Leaf Diseases

Abstract

Background

In this paper, the ensemble average and deep neural networks approach is proposed for detecting rice leaf diseases. The system consists of two different deep neural networks represented by GoogLeNet and MobileNetV2 where trained separately by dataset of rice leaves images.

Materials and Methods

 These images represent various conditions of the leaves, including healthy leaves, and five types of diseases. The dataset is preprocessed by resizing and normalizing to standardize the inputs for the networks. Then it is split into training and testing sets to ensure robust model evaluation. After training of these two networks, the ensemble average module combines the predictions from both networks by averaging them during the testing phase.

Results

The proposed model was compared with the other two models based on the performance metrics accuracy, precision, recall, and F1-score, and the proposed model highlights the superiority of the proposed approach in detecting rice leaf diseases. The proposed model achieved the highest accuracy at 97.07%, precision at 97.1%, recall at 97.06%, and F1-score at 97.08% which outperformed the other two models across all metrics because the averaging process reduces variance and enhances the accuracy of the final decision of detection the rice leaf diseases

Conclusion

This ensemble-based approach illustrates superior performance for rice leaf diseases detection, offering a more reliable and precise tool for managing the agricultural diseases.