With　the　acceleration　of　the　global　aging　process,　the　safety　and　health　of　the　elderly　has　become　a　widespread　concern,　and　falls　have　become　the　main　health　threat　of　the　elderly.　In　this　paper,　a　fall　detection　model　based　on　OpenPose　human　posture　estimation　algorithm　is　proposed　by　using　the　fall　detection　method　based　on　machine　vision.　On　the　basis　of　OpenPose　human　key　point　detection,　combined　with　SSD　MobileNet　object　detection　framework　can　remove　the　non　human　key　points　detected　by　OpenPose　algorithm,　reduce　the　false　detection　rate　of　the　algorithm,　improve　the　robustness　of　the　algorithm　in　complex　environment,　and　then　extract　the　features　of　human　joint　points,　use　SVDD　classification　algorithm　to　classify,　experiments　show　that　this　method　can　effectively　detect　the　occurrence　of　falls,　and　the　accuracy　rate　can　reach　more　than　93%.　©　2020　IEEE.