Counting Corn Seedlings from UAV-Based RGB Images Using Computer Vision and Deep Learning Methods in Significant Weed Infestation and Partial Occlusion Conditions

A method is proposed for automatically counting corn seedlings under significant weed infestation and partial occlusion using RGB images of ultrahigh spatial resolution ( <0.5 cm/pixel) acquired from an unmanned aerial vehicle. The method combines computer vision and machine learning algorithms. Experimental studies have shown that the accuracy of estimated average number of corn seedlings in the early stages of vegetation is ~97%