In the 21st century, with the rapid development of artificial intelligence (AI) technology, rapid and automatic recognition of ground objects in remote sensing images has gradually become the current research focus. This technology solves the problem that a single remote sensing image can not cover the target area and can not meet the actual operation requirements. Because of the advantages of high efficiency, low cost and reliable mosaic image quality, it is widely used in engineering surveying and mapping, traffic monitoring, military investigation, disaster investigation and other fields. With the rapid development of aerial remote sensing technology, the number of optical remote sensing images has increased explosively, which provides necessary data support for using deep learning method to detect targets in remote sensing images. Based on the research background of automatic detection and recognition of bridge targets in visible light remote sensing images, this paper studies the problems of feature extraction, river detection, river region correction and bridge target recognition, and constructs a complete automatic detection and location system of bridge targets. After that, the bridge is accurately detected using binary morphology combined with the feature of bridge crossing both banks.