Due　to　the　complex　background　and　spatial　distribution,　it　brings　great　challenge　to　object　detection　in　high-resolution　remote　sensing　images.　In　view　of　the　characteristics　of　various　scales,　arbitrary　orientations,　shape　variations,　and　dense　arrangement,　a　multiscale　object　detection　method　in　high-resolution　remote　sensing　images　is　proposed　by　using　rotation　invariance　deep　features　driven　by　channel　attention.　First,　a　channel　attention　module　is　added　to　our　feature　fusion　and　scaling-based　single　shot　detector　(FS-SSD)　to　strengthen　the　long-term　semantic　dependence　between　objects　for　improving　the　discriminative　ability　of　the　deep　features.　Then,　an　oriented　response　convolution　is　followed　to　generate　feature　maps　with　orientation　channels　to　produce　rotation　invariant　deep　features.　Finally,　multiscale　objects　are　predicted　in　a　high-resolution　remote　sensing　image　by　fusing　various　scale　feature　maps　with　multiscale　feature　module　in　FS-SSD.　Five　experiments　are　conducted　on　NWPU　VHR-10　dataset　and　achieve　better　detection　performance　compared　with　the　state-of-the-art　methods.