[Objective] To summarize remote sensing monitoring techniques used in dynamic monitoring project of soil and water loss in Yellow River basin in order to explore a new and efficient approach for soil and water loss monitoring.[Methods] Based on resource 3 high-resolution satellite images, the object-oriented image classification approach was used to make semi-automatic classification of land utilization in Jungar Banner in 2014. The vegetation regression models were constructed to study the inversion of vegetation coverage in the project area.[Results] Based on semi-automatic classification of land utilization and the regression model of vegetation coverage, the experiment showed that the accuracy of field verification reached more than 90%, which satisfied the demand of high resolution and accuracy data acquisition for soil erosion monitoring.[Conclusion] The object-oriented image classification of land utilization and regression model of vegetation coverage can not only effectively avoid the error caused by artificial visual interpretation and save cost, but also improve the accuracy and efficiency for data acquisition.