Early Detection of Missing Plants in Maize Crops Through UAV Imaging

The number of plants born after corn seed planting determines crop yields for farmers. Manual monitoring of missing corn plants requires resources and time to cover large areas of crops. Traditional crop monitoring and tracking methods can be replaced by unmanned aerial vehicles (UAVs) and systematized precision agriculture (PA) methods to count corn plants using offline imagery. The present research proposes a new algorithm for detecting missing plants in the first weeks of corn growth. The algorithm was developed in Matlab using computer vision to detect missing corn plants using RGB images captured by drones with heights of 5, 10, and 15 m. The experimentation was carried out with 30 images of each height captured in the third week of crop growth. The most appropriate height for better detection was established after an evaluation procedure with the set of images (90 in total). The evaluated algorithm obtained an accuracy of 80% with images of 5 m, 67% accuracy with images of 10 m, and 52% with images of 15 m height.