In addressing the issue of accuracy in large-scale spatial land use type classification based on remote sensing, precise and efficient land use classification extraction methods and land classification models suitable for diverse high-altitude mountainous terrains were compared and proposed. The random forest (RF) and support vector machine (SVM) algorithms were employed to classify land use in Yunnan Province. Accuracy validation was carried out by visually interpreting and randomly sampling 1 525 sample points. The results indicate that the application of RF and SVM classification algorithms both yield land use classification accuracy of over 80% in Yunnan Province. Cultivated land in Yunnan Province shows a trend of initial increase and subsequent decrease from 2019 to 2021. Using RF and SVM, overall accuracy and Kappa coefficients are effective for comparative analysis of land use classification. The RF algorithm demonstrates higher accuracy in identifying land features within the study area compared to SVM, making it more suitable for land use classification research in Yunnan Province's high-altitude mountainous terrain.