Leaf Images Assisted Potato Plant Disease Detection Using Deep-Learning with Features Fusion

Plant condition monitoring is one of the necessary tasks in the agriculture to confirm the yield. Recent agricultural monitoring procedures employed computerised-algorithms to automate several agricultural practices including the disease monitoring and alert. This research developed a deep-learning (DL) tool to detect the disease based on the plant leaf data and this work considered the potato plant for the demonstration. The developed procedure consist several stages; leaf data collection and processing, leaf feature extraction with MobileNet (MN) model, identification of best two model and getting best feature vector after 50% feature dropout, implementing features-fusion (FF) using selected model features, and detection using 3-fold cross-validation. The performance of this tool is verified using FF and machine-learning classifiers and this work verifies that the decision tree based detection offered an accuracy of 95%. This work confirms the significance of the developed DL-tool on the potato plant data.