Clouds are the core of weather observation and forecasting. They help promote the water cycle and the entire climate system. Accurate and automatic recognition of cloud types is one of the important applied researches to improve the accuracy of weather forecasts. At present, the recognition of the cloud shape of ground-based cloud images is a major and difficult problem in realizing the automation of ground cloud measurement instruments. Therefore, this paper builds a neural network framework training model based on wireless sensors and Python+TensorFlow, and develops a ground-based cloud image cloud recognition system. After experiments, the system can run successfully, and the experiments on the cloud map data set published by Guanyun Zhitian show that the recognition accuracy of the system can reach about 87%.
Software And Hardware