Extraction of Sunflower Lodging Information Based on UAV Multi-Spectral Remote Sensing and Deep Learning

The rapid and accurate identification of sunflower lodging is important for the assessment damage to crops. To develop a fast method extraction information on lodging, this study improves inputs SegNet U-Net render them suitable multi-band image processing. Random forest two improved deep learning methods are combined with RGB, RGB + NIR, red-edge, NIR red-edge bands multi-spectral images captured by UAV (unmanned aerial vehicle) construct 12 models extract lodging. These then used ignore edge-related predict results experiments show that were superior random in terms obtained accuracy. predictive accuracy model constructed using combination had highest overall 88.23%. Adding whereas adding reduced it. An overlay analysis area shows error was mainly caused failure recognize mixed areas low-coverage areas. when ignored about 2% higher than direct splicing method.