Inability of spatial transformations of CNN feature maps to support invariant recognition

Open in new window