When　a　space　heating　ASHP　unit　is　operated　in　frosting　conditions,　accurate　defrosting　is　necessary　to　ensure　its　operating　efficiency　and　output　heating　capacity.　Therefore,　to　improve　the　accuracy　of　defrosting　initiation　of　ASHPs　under　frosting-defrosting　operations,　a　study　on　applying　image　recog-nition　to　frost　built-up　detection　in　ASHPs　has　been　carried　out　and　the　study　results　are　presented　in　this　paper.　Firstly,　the　principle　of　a　novel　frost　built-up　detection　method　based　on　image　recognition　is　presented.　Secondly,　a　series　of　experiments　to　examine　the　applicability　of　the　novel　frost　built-up　detection　method　with　respect　to　shooting　angle,　imaging　pixels,　illumination　level,　and　the　tempera　-ture　of　outdoor　coil　surface　are　carried　out.　Thirdly,　the　novel　frost　built-up　detection　method　was　further　modified　based　on　the　experimental　result　obtained,　and　the　verification　of　the　modified　novel　detection　method　is　reported.　It　is　shown　that　the　use　of　the　modified　novel　frost　built-up　detection　method　could　properly　account　for　the　impact　of　changing　illuminance　level　on　the　accuracy　of　image　recognition,　to　ensure　the　accuracy　of　frost　built-up　detection　when　ASHPs　are　operated　in　frosting　conditions.　(C)　2021　Elsevier　Ltd.　All　rights　reserved.