Research on Multi-class Weather Classification Algorithm Based on Multi-model Fusion
Yin Wang, Yingxiang Li
Abstract
In daily life, the animal husbandry, aquaculture, agriculture, and transportation industries that people engage in are all affected by the weather. There is no doubt that accurate, convenient and fast identification of various types of weather is of great significance in production and life. In this article, we collected and labeled a dataset containing 9 types of ground weather images. In addition, we use ResNet convolutional neural network and DenseNet convolutional neural network to build the network structure, and perform probability discrimination on the output results of each model to increase the recognition rate. We trained and tested on the collected weather dataset. Experimental results show that the method has better performance in weather recognition.