A　method　that　uses　machine　vision　and　machine　learning　technologies　to　identify　the　endhead　in　a　steel　coil　has　seldom　been　proposed.　In　this　study,　an　improved　faster　region-based　convolutional　neural　network　(F-RCNN)　deep　learning　algorithm　is　introduced　to　identify　the　position　of　the　steel　coil　end-head　for　a　hardware　system　set　up　for　image　sensing　and　detection.　The　feature　pyramid　network　(FPN)　and　the　parallel　attention　module　(PAM),　which　are　both　involved　in　the　traditional　F-RCNN,　are　used　to　increase　the　detection　accuracy.　Our　experimental　results　validated　the　effectiveness　of　the　proposed　improved　algorithm.