The evergreen broad-leaved forest is a zonal vegetation type in the subtropical regions of China. However, due to complex vertical structure and relatively dense canopy of subtropical forests, we are lack of reliable and long-term monitoring data in current canopy studies of this forest. Here we present a large dataset of visible light remote sensing images of four large-scale (> 20 ha) forest dynamic plots in Tiantongshan and Baishanzu of Zhejiang, and Dinghushan and Chebaling of Guangdong. These data were collected by unmanned aerial vehicle (UAV) equipped with visible light camera in 2014-2016. By combining the UAV data with ground survey data, we included three data files for each plot: orthoimages and surface digital models with spatial resolution of ~5 cm, and forest canopy height data with spatial resolution of 1 m. This dataset of evergreen broad-leaved forest provides significant supports for the studies in forest canopy ecology, biodiversity and ecosystem function.