Accurate　identification　of　impact　loads　is　vital　for　structural　assessment　and　design.　Traditional　methods　rely　on　complex　equipment,　such　as　accelerometers　or　strain　gauge,　which　can　be　expensive　and　prone　to　failure.　This　study　introduces　a　non-contact　intelligent　identification　approach　incorporating　visual　sensing　technology,　providing　a　convenient　means　to　identify　impact　loads.　Numerical　simulations　explore　the　differences　in　identifying　impact　forces　through　acceleration　and　displacement　responses,　particularly　by　considering　such　variables　as　measurement　noise　and　number　of　measurement　points.　A　meticulously　designed　experiment　verified　the　feasibility　of　the　proposed　method　for　measuring　the　displacement　and　velocity　of　rapidly　moving　targets,　and　evaluated　its　performance　in　terms　of　accuracy.　A　series　of　impact　loading　experiments　were　conducted　on　a　simply　supported　girder　bridge　model　to　validate　the　effectiveness　of　the　proposed　impact　force　identification　method.　Results　indicate　strong　agreement　between　displacement　response　measurements　and　percentile　meters.　The　proposed　non-contact　method　accurately　identifies　single　or　continuous　impact　loads,　with　a　minimum　peak　relative　error　of　0.2%.　This　study　represents　a　pioneering　application　of　intelligent　visual　sensing　technology　in　the　field　of　impact　load　identification.　Moreover,　the　current　research　introduces　a　novel　approach　to　address　the　challenges　faced　by　conventional　methods　in　identifying　impact　loads.　Future　research　can　leverage　the　groundwork　laid　by　this　study　to　further　optimize　and　expand　the　proposed　method,　enhancing　its　capabilities　and　fully　harnessing　its　potential　to　offer　advanced　solutions　in　structural　health　monitoring.