The accurate stand count determination in corn field at the early-season is of great interest to corn breeders and plant geneticists. But the most common method is counting the number of plants manually, which is time-consuming, laborious and prone to be error. In this research project we developed an automatic, robust and high-throughput method for reliable corn stand counting based on color images extracted from video clips. Detecting corn stands in the field is a challenging task primarily because of, the camera motion, leaf fluttering caused by wind, shadows of plants caused by direct sunlight, and the complex soil background. Especially, the detection of early corn seedings at around one week after planting, where the plants are small and difficult to differentiate from the background. We developed a pipeline based on YOLOv3 network that was enhanced by image pre-processing team. The result demonstrated that our method is accurate and reliable for stand counting.
