# DualFormer: Local-Global Stratified Transformer for Efficient Video Recognition

arXiv: https://arxiv.org/abs/2112.04674 · HTML: https://mlsanity.com/paper/2112.04674/

## Key facts

- Cited by 11 papers in the map (data through 14 Oct 2026).
- Builds on 54 of its 66 references that are in the map.
- First posted on arXiv 9 Dec 2021.

## Builds on (most cited first, 30 of 54)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Very Deep Convolutional Networks for Large-Scale Image Recognition](https://mlsanity.com/paper/1409.1556/) (2014)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization](https://mlsanity.com/paper/1610.02391/) (2016)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://mlsanity.com/paper/2010.04159/) (2020)
- [Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset](https://mlsanity.com/paper/1705.07750/) (2017)
- [Xception: Deep Learning with Depthwise Separable Convolutions](https://mlsanity.com/paper/1610.02357/) (2016)
- [Non-local Neural Networks](https://mlsanity.com/paper/1711.07971/) (2017)
- [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://mlsanity.com/paper/2102.12122/) (2021)
- [The Kinetics Human Action Video Dataset](https://mlsanity.com/paper/1705.06950/) (2017)
- [SlowFast Networks for Video Recognition](https://mlsanity.com/paper/1812.03982/) (2018)
- [Vision Transformers for Dense Prediction](https://mlsanity.com/paper/2103.13413/) (2021)
- [Is Space-Time Attention All You Need for Video Understanding?](https://mlsanity.com/paper/2102.05095/) (2021)
- [ViViT: A Video Vision Transformer](https://mlsanity.com/paper/2103.15691/) (2021)
- [A Closer Look at Spatiotemporal Convolutions for Action Recognition](https://mlsanity.com/paper/1711.11248/) (2017)
- [The "something something" video database for learning and evaluating visual common sense](https://mlsanity.com/paper/1706.04261/) (2017)
- [Video Swin Transformer](https://mlsanity.com/paper/2106.13230/) (2021)
- [Deep Networks with Stochastic Depth](https://mlsanity.com/paper/1603.09382/) (2016)
- [Multiscale Vision Transformers](https://mlsanity.com/paper/2104.11227/) (2021)
- [TSM: Temporal Shift Module for Efficient Video Understanding](https://mlsanity.com/paper/1811.08383/) (2018)
- [Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition](https://mlsanity.com/paper/1406.4729/) (2014)
- [X3D: Expanding Architectures for Efficient Video Recognition](https://mlsanity.com/paper/2004.04730/) (2020)
- [Network In Network](https://mlsanity.com/paper/1312.4400/) (2013)
- [Twins: Revisiting the Design of Spatial Attention in Vision Transformers](https://mlsanity.com/paper/2104.13840/) (2021)
- [Can Spatiotemporal 3D CNNs Retrace the History of 2D CNNs and ImageNet?](https://mlsanity.com/paper/1711.09577/) (2017)

## Built on by (most cited first, 11 of 11)

- [Mugs: A Multi-Granular Self-Supervised Learning Framework](https://mlsanity.com/paper/2203.14415/) (2022)
- [STAR-Transformer: A Spatio-temporal Cross Attention Transformer for Human Action Recognition](https://mlsanity.com/paper/2210.07503/) (2022)
- [An Efficient Spatio-Temporal Pyramid Transformer for Action Detection](https://mlsanity.com/paper/2207.10448/) (2022)
- [Learning Correlation Structures for Vision Transformers](https://mlsanity.com/paper/2404.03924/) (2024)
- [X-Light: Cross-City Traffic Signal Control Using Transformer on Transformer as Meta Multi-Agent Reinforcement Learner](https://mlsanity.com/paper/2404.12090/) (2024)
- [Linear Video Transformer with Feature Fixation](https://mlsanity.com/paper/2210.08164/) (2022)
- [Transformer in Transformer as Backbone for Deep Reinforcement Learning](https://mlsanity.com/paper/2212.14538/) (2022)
- [PDiT: Interleaving Perception and Decision-making Transformers for Deep Reinforcement Learning](https://mlsanity.com/paper/2312.15863/) (2023)
- [TRIFFID: Autonomous Robotic Aid For Increasing First Responders Efficiency](https://mlsanity.com/paper/2502.09379/) (2025)
- [DVFL-Net: A Lightweight Distilled Video Focal Modulation Network for Spatio-Temporal Action Recognition](https://mlsanity.com/paper/2507.12426/) (2025)
- [Efficient Selective Audio Masked Multimodal Bottleneck Transformer for Audio-Video Classification](https://mlsanity.com/paper/2401.04154/) (2024)

## Abstract (arXiv)

While transformers have shown great potential on video recognition with their strong capability of capturing long-range dependencies, they often suffer high computational costs induced by the self-attention to the huge number of 3D tokens. In this paper, we present a new transformer architecture termed DualFormer, which can efficiently perform space-time attention for video recognition. Concretely, DualFormer stratifies the full space-time attention into dual cascaded levels, i.e., to first learn fine-grained local interactions among nearby 3D tokens, and then to capture coarse-grained global dependencies between the query token and global pyramid contexts. Different from existing methods that apply space-time factorization or restrict attention computations within local windows for improving efficiency, our local-global stratification strategy can well capture both short- and long-range spatiotemporal dependencies, and meanwhile greatly reduces the number of keys and values in attention computation to boost efficiency. Experimental results verify the superiority of DualFormer on five video benchmarks against existing methods. In particular, DualFormer achieves 82.9%/85.2% top-1 accuracy on Kinetics-400/600 with ~1000G inference FLOPs which is at least 3.2x fewer than existing methods with similar performance. We have released the source code at https://github.com/sail-sg/dualformer.