Image description through fusion based recurrent multi-modal learning | IEEE Conference Publication | IEEE Xplore