Insulators　are　the　most　common　equipment　in　the　power　system,　the　failure　of　insulators　will　cause　heavy　economic　loss　to　electric　power　companies,　so　it　is　very　important　to　detect　insulators　effectively　for　inspecting　their　working　states.　This　paper　proposes　a　novel　method　to　detect　the　insulators　based　on　Faster　R-CNN　in　which　Region　Proposal　Network　(RPN)　is　used　to　generate　high-quality　insulator　candidates　and　the　convolution　features　are　shared　with　Fast　R-CNN　to　detect　the　insulator.　A　large　number　of　visible　light　images　are　used　as　experimental　data　in　experiment,　and　the　results　show　that　this　method　can　detect　insulators　in　complex　background　with　high　precision　as　well　as　low　time　cost.　©　2017　IEEE.