A　license　plate　detection　and　recognition　system　is　one　of　the　practical　applications　of　computer　vision　technology　in　the　field　of　unmanned　vehicles.　In　this　paper,　we　proposed　a　Light-yolov7　for　license　plate　detection　and　recognition　model,　which　is　applied　to　unmanned　vehicles.　The　model　contains　three　improvements:　a　lightweight　neural　network　ShuffleNet2　is　used　for　feature　extraction,　a　depth-separable　convolution　is　added　to　reduce　the　number　of　parameters,　then　this　paper　uses　late　fusion　to　connect　features.　Finally,　CRNN　is　used　to　learn　the　obtained　features.　Experiments　on　a　large　Chinese　license　plate　dataset　(CCPD+CRPD)　show　that　the　model　is　feasible　for　mobile　deployment　and　efficient　for　license　plate　detection　and　recognition.　©　2023,　The　Author(s),　under　exclusive　license　to　Springer　Nature　Singapore　Pte　Ltd.