Ensemble Learning for Weed Detection in Soybean Crop

In precision agriculture, to reduce the herbicides and pesticides usage for good crop's yield, efficient weed management measures is necessary. This research explores the application of ensemble deep learning techniques for weed detection in soybean fields, aiming to enhance precision agriculture practices. The ensemble approach combines the strengths of individual models, providing a robust and accurate solution to the complex task of weed detection. The experiments are done on publicly available soybean and weed dataset containing soybean, grass, broadleaf and soil images. For preprocessing of images CLAHE algorithm is used. Different pre-trained deep learning networks viz., Darknet19, Mobilenetv2, VGG19, Resnet18, Inceptionv3 and Densenet201 are investigated for classification of weed/crop. Also an ensemble model is designed with top three accurately classified pre trained networks viz., Resnet18, Inceptionv3 and Densenet201. Ensemble method yielded an accuracy of 99.72%, which is better than base models. Extensive experiment results reveal that the ensemble technique is better than individual models and its ability to reduce false positives and negatives.