Learning cross-modal appearance models with application to tracking | IEEE Conference Publication | IEEE Xplore