This　paper　presents　a　parallel　face　detection　system　based　on　multi-core　processors,　of　which　the　hardware　part　uses　the　ZYNQ　chips　and　multi-core　processors　and　the　software　part　includes　the　input　and　output　of　the　image　data　and　parallel　detection.　The　part　of　parallel　detection　uses　the　cascade　classifier　based　on　MB-LBP　feature　and　Gentle　Adaboost　algorithm.　This　system　is　aimed　for　a　low-power　multicore　processor　that　is　well　suited　to　embedded　system.　The　experimental　results　indicate　that　the　parallel　face　detection　system　reduces　the　running　time　of　the　algorithm　comparing　with　that　on　the　ARM　platform.　Moreover,　our　system　has　outstanding　performance　and　very　high　accuracy.　©　2017　IEEE.