Deriving high-level scene descriptions from deep scene CNN features | IEEE Conference Publication | IEEE Xplore