UAV Images Based Maize Crop Analysis Using Custom CNN Model

Precision agriculture is undergoing revolution by the integration of modern technologies such as Internet of Things (IoT), Unmanned Aerial Vehicles (UAVs), Artificial Intelligence (AI) and Machine Learning (ML). In this paper, high resolution UAV images of maize crops is used and further split into low resolution sub-images for training Convolutional Neural Network (CNN) models. The paper work provides an practical approach of applying an lightweight custom CNN architecture on UAV images to analyse the maize crops health status by classifying them into healthy crops, Maydis Leaf Blight (MLB) disease crops and also weeds and soil back ground images. The proposed custom CNN model shows better performance by achieving 95% accuracy when compared to other considered CNN pre-trained models such as MobileNet-V2 and ResNet-50.