Twins: Revisiting the Design of Spatial Attention in Vision Transformers

Very recently, a variety of vision transformer architectures for dense prediction tasks have been proposed and they show that the design of spatial attention is critical to their success in these tasks. In this work, we revisit the design of the spatial attention and demonstrate that a carefully-devised yet simple spatial attention mechanism performs favourably against the state-of-the-art schemes. As a result, we propose two vision transformer architectures, namely, Twins-PCPVT and Twins-SVT. Our proposed architectures are highly-efficient and easy to implement, only involving matrix multiplications that are highly optimized in modern deep learning frameworks. More importantly, the proposed architectures achieve excellent performance on a wide range of visual tasks, including image level classification as well as dense detection and segmentation. The simplicity and strong performance suggest that our proposed architectures may serve as stronger backbones for many vision tasks. Our code is released at .

PDF Abstract NeurIPS 2021 PDF NeurIPS 2021 Abstract

Results from the Paper

Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Semantic Segmentation ADE20K Twins-SVT-L (UperNet, ImageNet-1k pretrain) Validation mIoU 50.2 # 110
Semantic Segmentation ADE20K val Twins-SVT-L (UperNet, ImageNet-1k pretrain) mIoU 50.2 # 48
Image Classification ImageNet Twins-SVT-L Top 1 Accuracy 83.7% # 369
Number of params 99.2M # 867
GFLOPs 15.1 # 341