An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
By Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Mostafa Dehghani, Matthias Minderer, Georg Heigold and 3 more. International Conference on Learning Representations, 2020
66,490 citations recorded by the source corpus.
While the Transformer architecture has become the de-facto standard for natural language processing tasks, its applications to computer vision remain limited. In vision, attention is either applied in conjunction with convolutional networks, or used to replace certain components of convolutional networks while keeping their overall structure in place. We show that this reliance on CNNs is not necessary and a pure transformer applied directly to sequences of image patches can perform very well on image classification tasks. When pre-trained on large amounts of data and transferred to multiple mid-sized or small image recognition benchmarks (ImageNet, CIFAR-100, VTAB, etc.), Vision Transformer (ViT) attains excellent results compared to state-of-the-art convolutional networks while requiring substantially fewer computational resources to train.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Model Architectures
View Paper: https://arxiv.org/abs/2010.11929
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Mohit Bansal
- Jianfeng Gao
- Quoc V. Le
- Yoshua Bengio
- Sergey Levine
- Percy Liang
- Luke S. Zettlemoyer
- Dacheng Tao
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Trevor Darrell
- Xiaogang Wang
- Jitendra Malik
- Andrew Zisserman
- Yi Yang
In Model Architectures
- Quoc V. Le
- Yoshua Bengio
- Jianfeng Gao
- Alan L. Yuille
- Jiashi Feng
- Dacheng Tao
- Christoph Feichtenhofer
- Kaiming He
Citations in the corpus
- Learning Transferable Visual Models From Natural Language Supervision (2021, methodology): “Vision Transformer (ViT) We also include four ViT (Dosovitskiy et al., 2020) checkpoints pretrained on the ImageNet-21k dataset, namely ViT-B/32, ViT-B/16, ViTL/16, and ViT-H/14.”
- Scale Efficiently: Insights from Pre-training and Fine-tuning Transformers (2021, methodology): “• To consider how generalized these scaling strategies are, we conduct additional experiments on Vision Transformers (ViT; Dosovitskiy et al., 2020) to verify them in the vision domain.”
- FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness (2022, methodology): “Transformer-based models have become the most widely-used architecture in natural language processing [24] and computer vision [26, 95].”
- CodeGen: An Open Large Language Model for Code with Multi-Turn Program Synthesis (2022, methodology): “, 2020), computer vision (Dosovitskiy et al., 2021), and many other areas (Oord et al.”
- Measuring the Carbon Intensity of AI in Cloud Instances (2022, methodology): “Vision Transformers.”
- A Review of Sparse Expert Models in Deep Learning (2022, background): “Some examples are image classification (Dosovitskiy et al., 2020), object detection (Carion et al., 2020), recommendation systems (Chen et al., 2019), speech recognition (Dong et al., 2018; Nakatani, 2019; Gulati et al., 2020).”
- Natural Adversarial Examples (2019, background): “We now assess performance on a completely different architecture which does not use convolutions, vision Transformers [12].”
- Benchmarking Detection Transfer Learning with Vision Transformers (2021, methodology): “…role in the study of transfer learning over the last decade ( e.g ., [35, 14, 9, 17]), was not explored in the pioneering work on ViT training [10, 7, 5]—supervised or unsupervised—due to the challenges (described shortly) of integrating ViTs into common detection models, like Mask R-CNN [19].”
- Early Convolutions Help Transformers See Better (2021, methodology): “Next, we review vision transformers [13] and describe the convolutional stems used in our work.”
- Towards Robust Vision Transformer (2021, neutral): “ViTs [10] tokenize an image by splitting it into patches with size of 16×16 or 32×32.”
- Emerging Properties in Self-Supervised Vision Transformers (2021, methodology): “Transformers [70] have recently emerged as an alternative to convolutional neural networks (convnets) for visual recognition [19, 69, 83].”
- An Empirical Study of Training Self-Supervised Vision Transformers (2021, methodology): “On one hand, this comparison shows the potential of ViT, especially considering that it achieves these results using relatively “fewer inductive biases” [15].”
- Self-supervised Pretraining of Visual Features in the Wild (2021, background): “[7] as well as the pretraining on curated data from SimCLRv2 [9] and ViT [14].”
- Training data-efficient image transformers & distillation through attention (2020, methodology): “By default and similar to ViT [15] we train DeiT models at resolution 224× 224 and we fine-tune at resolution 384×384.”