利用SAM模型的地理信息更新关键技术研究OA

With the increasing abundance of remote sensing image data,traditional human-computer interactive visual interpretation and deep learning model interpretation are facing numerous challenges in the update of geographical information.Focusing on the actual demand for rapid and accurate update of geographical information,we introduced the powerful image deconstruction capability of SAM model,combined technolo-gies such as the interactive segmentation framework of multi-modal prompts and active contour models to optimize and fine-tune the SAM mod-el,and proposed a geographical information patch update technology based on SAM model.Experimental results show that this model demon-strates higher accuracy and robustness in handling the tasks of geographical patch segmentation and interpretation,can significantly improve the automation in geographical data processing and mapping accuracy,and help promote the transformation of surveying,mapping and geographical information work from the traditional manual mode to an intelligent and efficient direction.