3D2SeqViews: Aggregating Sequential Views for 3D Global Feature Learning by CNN With Hierarchical Attention Aggregation | IEEE Journals & Magazine | IEEE Xplore