Abstract:[Objective] To investigate the spatio-temporal change characteristics and driving mechanisms of understory vegetation cover, [Methods] Hubei Province was selected as the study area, and understory green leaf vegetation cover and understory litter cover were extracted from 28 sample plots for every half month in 2022 through field survey and DeepLabV3+ semantic segmentation method. Based on this, various machine learning models were used to analyze the impact of four driving factors, including spatial location, natural environment, social and economic environment, and climate conditions, on the change of understory green leaf vegetation cover. [Results] The results showed that the monthly variation of understory litter cover did not show obvious seasonal characteristics, and there was considerable spatial and temporal heterogeneity in understory litter cover among the sample plots. In contrast, understory green leaf vegetation cover showed obvious seasonal change characteristics, and there were significant differences in understory green leaf vegetation cover among different vegetation types: the understory green leaf vegetation cover of economic forests and conifers was generally higher than that of broadleaf deciduous forests and evergreen broadleaf forests, and the differences of the understory green leaf vegetation cover between broadleaf deciduous forests and evergreen broadleaf forests were relatively small. The random forest regression model had the best prediction performance, with a root mean square error (RMSE) of 11.0723 and a coefficient of determination (R2) of 0.732. [Conclusion] The random forest regression model shows that the temperature, NDVI, and precipitation in the previous month are the key driving factors for the spatio-temporal change of understory green leaf vegetation cover. The results of this study provide scientific evidence for the management of understory vegetation and ecological protection.