2022-11-03 Vision TransformersSegNeXt: Rethinking Convolutional Attention Design for Semantic Segmentation