Building construction green dust prevention net identification method and system, computer device and readable storage medium

Abstract

The building construction green dust prevention net identification method, system, computer device and readable storage medium belong to the dust source identification field, and aim to solve the problems of low efficiency, poor precision, missing boundary information and the like of the existing method. The method first acquires and pre-processes a city built area remote sensing image, generates a grid label, divides a training set and a verification set and enhances data; then trains a DMS-UNet model and a comparison model, and screens an optimal model; then adopts an edge prediction ignoring algorithm to segment and identify a new image, generates spatial distribution data and analyzes the influence. The DMS-UNet model is additionally provided with three modules of deformable convolution, multi-scale feature pyramid and dynamic upsampling, so that the deformation adaptation and feature capture capability are improved. The system contains multiple functional modules in cooperation, the device contains a memory and a processor, and the storage medium stores a program. The present application realizes high-precision and efficient identification of the green dust prevention net, provides decision support for the environmental protection department to accurately locate and control the dust source, and helps the urban environmental governance.

Images