Corn Plants Disease Detection and Prediction using Deep Ensemble Model

Agriculture plays an important role in a country, and agriculture also contributes greatly to the survival of mankind. It is important to obtain pesticide-cleaning crops to improve yields in agriculture. If there are pest-rich crops, it produces good yields and increases the country’s income. The research is now based on IT-based technologies to detect infectious diseases in crops. To do so, an ensemble model is tested. The Ensemble model combines three deep learning model with EfficientDet, Faster Region-based Convolutional Neural Network (Faster R-CNN) and Single-Shot-Detector (SSD). Research combines three deep learning models and ensemble models to detect and estimate diseases that fall in corn leaf using images of corn obtained from the Plant Village data set. Using the Ensemble Model, the proposed system will select the bounding box detected by the three models. And then, the system will select the suitable bounding box using Non-Maximum Suppression (NMS). Therefore, testing with an ensemble model that combines three models provides a more accurate result than searching for and predicting microbes that can fall in corn using each Deep learning model. With the Ensemble model, more than ten thousand training images and more than three thousand testing images, mAP Score receives 95.00. The result is acceptable in finding and estimating insects that fall in plants.