Fine urban land use type identification model training method

The invention discloses a fine urban land use type identification model training method, which comprises the following steps: dividing a land parcel area into a plurality of grids arranged in grids, taking the land parcel area as a land use identification window, taking a central grid of the land use identification window as a to-be-identified land parcel, and taking a grid adjacent to the central grid as an adjacent land parcel; obtaining feature data of each grid, wherein the feature data comprises a point-of-interest type, a land type, a population density type and whether the grid is in a central urban area; selecting a land parcel with a known land type as a central land parcel of the land identification window, taking a land parcel adjacent to the central land parcel as a sample, taking the feature data of the sample as a feature set, and constructing a sample feature data set; and taking the sample feature data set as model input, taking the sample tag data set as model output, and ...