Research on small building detection method based on convolutional neural network

Abstract

As one of the main targets of remote sensing technology detection, how to automatically and efficiently extract small buildings from large quantities of data is the main problem of urban planning and construction. At present, although domestic and foreign scholars have made excellent achievements in the research of remote sensing technology, target detection and recognition and extraction cannot fully meet the actual needs because of the inconsistent structure types of small buildings and the complex environment. Therefore, on the basis of understanding the research status of building extraction and convolutional neural network, this paper compares and analyzes the traditional detection algorithm and the semantic segmentation method based on convolutional neural network. The final results show that the extraction results of small buildings are more accurate and meet the needs of current urban planning and construction.