Sunflower Seedling Counting and Weed Detection by Deep Learning on Unmanned Aerial Vehicle Captured RGB Images

Some automated methods of sunflower seedling counting and weed vegetation mapping by unmanned aerial vehicle captured RGB images of ultrahigh spatial resolution are proposed. The methods are based on the application of convolutional DeepLabv3+, ResNet-101, and YOLOv8 neural networks. The results of experimental studies show that the sunflower seedling counting precision at early stages is averagingly 96 $$%$$ .