Plastic　greenhouse　is　a　widely　used　cultivation　facility　in　agricultural　production,　which　can　effectively　reduce　disasters　and　increase　production.　Accurate　acquisition　of　greenhouse　coverage　information　provides　scientific　data　support　for　optimizing　agricultural　production　layout　and　promoting　the　development　of　facility　agriculture.　We　obtained　Sentinel-2　image　data　of　Fujian　Province　in　the　spring,　autumn,　and　winter　from　2019　to　2020　through　Google　Earth　Engine　(GEE).　We　extracted　a　number　of　spectral,　textural,　and　DEM　features,　and　used　the　Gini　index　to　conduct　feature　extraction.　Finally,　we　used　the　Random　Forest　(RF)　classifier　to　extract　agricultural　greenhouses　in　different　seasons　in　Fujian　Province.　The　results　show　that:　(1)　Combined　with　GEE　and　RF,　the　overall　extraction　accuracy　of　Fujian　Province　is　84%,　implying　that　this　method　is　suitable　for　large-scale　greenhouse　extraction.　Among　them,　texture　features　have　a　significant　impact　on　the　classification　of　greenhouses.　(2)　Greenhouses　are　mainly　distributed　in　Minnan　Plain　of　Fujian　Province,　accounting　for　45.88%　of　the　total　area　of　Fujian　Province,　but　less　in　inland　areas.　(3)　Comparing　the　area　of　greenhouses　in　different　seasons,　it　was　found　that　the　largest　area　was　covered　by　greenhouses　in　spring,　followed　by　winter　and　autumn.　©　2022　IEEE.