High temperature and drought are the main stress sources affecting crop growth and final productivity. At present, UAV remote sensing technology has made great progress in the hierarchical monitoring of crop lodging and pests and diseases, but there are few reports on the use of UAV remote sensing for crop drought resistance grade monitoring. Therefore, taking ramie germplasm resources as the research object, quantitative criteria for ramie drought resistance was proposed, and a method to identify the drought resistance of ramie germplasm resources was providedby multi-spectral remote sensing of UAV. Firstly, totally 36 ramie germplasm resources were graded for drought resistance by experts. Then, combined with the vegetation index obtained by UAV multi-spectral remote sensing, and three machine learning methods,random forest (RF), support vector machine (SVM) and decision tree (DT) were used to construct ramie drought resistance identification models, and the results were evaluated by testing the phenotypic response of ramie under high temperature and drought stress. Finally, high-quality ramie germplasm resources under high temperature and drought stress were screened based on the remote sensing phenotypes obtained by UAV. The results showed that the accuracy of the ramie drought resistance identification model constructed by SVM reached 0.74, and the F1-score of different drought resistance classes was ranged from 0.69 to 0.79, indicating that the method could be used to evaluate the drought resistance of ramie germplasm resources. Three phenotypic characters of ramie (SPAD value, leaf area index and plant height) obtained from UAV remote sensing data were strongly correlated with the measured values. On this basis, three high-quality ramie germplasm resources PJ-CD, WS-XM and Xiangzhu 7 were selected from high temperature and drought stress.