Typical forage image recognition method based on U-Net network

In view of the low degree of monitoring and digitization of desertification grassland in Inner Mongolia, this paper constructs an improved network model based on U-Net network and Mobile network based on using unmanned air vehicle to obtain images of desertification grassland. The typical forage grasses stipa longiflora, kochia prostrate, alfalfa, vicia costata and agropyron cristatum in decertified grassland were classified and distinguish. The experimental results show that the improved U-Net network model can effectively realize the classification and identification of forage grass. The Kappa value of the network model is 84.8%, the overall accuracy is 89.9%, and the classification result is reliable. It can provide a scientific basis for monitoring vegetation species diversity, grass species degradation and pest control.