Hot-NetVLAD: Learning Discriminatory Key Points for Visual Place Recognition | IEEE Journals & Magazine | IEEE Xplore