Facility agricultural land multi-category identification and classification method based on remote sensing image

The invention belongs to the field of facility agricultural land distribution pattern research and optimization, particularly relates to a remote sensing image-based facility agricultural land multi-class identification and classification method, and solves the problem of low facility agricultural land identification precision of an existing neural network. The method is realized based on a semantic segmentation neural network, and the semantic segmentation neural network comprises an initialization module, an Encoder module, a pyramid pooling module and a coding module which are arranged in sequence. By utilizing the network, classification and positioning of the facility agricultural land are carried out on the remote sensing image, classification and positioning with the accuracy rate of more than 65% of multi-classification results and the accuracy rate of more than 90% of single-classification results are realized, manpower and material resources required by field investigation of the facility agricultur