Solar energy is the most common renewable energy source that is clean, safe, and inexhaustible. Photovoltaic (PV) technology utilizes solar panels to convert solar energy into electricity. The number of PV installations has rapidly increased worldwide. In this study, we employed the random forest classifier to extracted PV installations throughout China in 2015 and 2020 using Landsat-8 imagery based on the Google Earth Engine platform. The classification results were further visually inspected and refined by morphological filtering, cavity filling as well as manual contour adjustment. Validation analysis revealed that the resulting dataset achieved an overall accuracy over 96% for both 2015 and 2020. "ChinaPV" contains the specific location as well as the size of each PV installation. This study generated two vectorized soloar PV installation maps in China for the year 2015 and 2020. It includes the location, size, and perimeter of each PV installation. It also shows which province is each PV installed and whether it is located in urban areas or rural areas. ChinaPV is delivered in “ESRI Shapefile” formats in WGS-84 coordinate system. The attributes table of the PV polygons include the PV polygon ID, latitude and longitude coordinates of the centre point in each PV installation, the area (km2) and perimeter (km) of each PV, the name of province this PV locates, as well as whether this PV is located in urban or rural areas. Users can employ our ChinaPV dataset in 1) analysing the spatial and temporal patterns of PV installation across China and in different administrative provinces; (2) analysing the spatial and temporal patterns of PV installation over different land cover and land use types; (3) collecting PV samples to train a deep learning model; (4) estimating the generated electricity and carbon mitigation effect from solar PV; (5) evaluating the environmental impact of PV on hydrology and local climate.