Abstract:The types of soil and water conservation measures and their configuration modes are complicated. Accurate identification and fine extraction of detailed configuration information of soil and water conservation measures are the basis for obtaining the factor values of soil and water conservation measures. The information acquisition methods of soil and water conservation measures mainly include traditional field surveys, satellite remote sensing images, and UAV close-range photography. The identification and extraction methods mainly include visual interpretation, traditional machine learning, object-oriented classification methods, and deep learning models. This paper summarizes the existing shortcomings and puts forward the research prospect by combing the research results of soil and water conservation measures identification and extraction methods at home and abroad. In semantic segmentation, future feature fusion and multimodal learning, weak supervision and semi-supervised learning, ensemble learning and meta-learning can be applied to the extraction of soil and water conservation measures. At present, there are few reports on the results of identification and extraction of soil and water conservation tillage measures, while tillage measures are common in agricultural practice, and the research on identification and extraction of tillage measures should be strengthened in the future. Artificial intelligence combined with big data technology is the development direction of efficient and accurate identification and extraction of soil and water conservation measures information in the future, and it is necessary to further study the semi-supervised and weakly supervised learning, semantic segmentation and instance segmentation methods, combined with multi-modal learning, small sample label and other methods to extract the information of soil and water conservation points, linear engineering measures, plant measures and tillage measures, so as to improve the information extraction methods of various soil and water conservation measures, and provide support for accurately obtaining the factor values of soil and water conservation measures and accounting the carbon sink capacity of soil and water conservation.