Methods › Computer Vision › Vision Transformers

Vision Transformers

42 methods 2,733 papers tagged archive 2025-07-28

The archive attaches this collection's text per method and the copies differ: 6 distinct texts across 42 of the 42 methods here. All are shown, most-carried first (a tie goes to the text carrying Papers with Code's collection boilerplate, then to the longer text); no vote is taken between them.

Text 1, carried by 37 of 42 methods:

Vision Transformers are Transformer-like models applied to visual tasks. They stem from the work of ViT which directly applied a Transformer architecture on non-overlapping medium-sized image patches for image classification. Below you can find a continually updating list of vision transformers.

According to [1], ViT type models can be further categorized into uniform scale ViTs, multi-scale ViT, hybrid ViTs with convolutions, and self-supervised ViTs. The methods listed below provide a comprehensive overview of ViT models applied to a range of vision tasks.

[1] Transformers in Vision: A Survey

Text 2, carried by 1 of 42 methods:

Self-Supervised Learning refers to a category of methods where we learn representations in a self-supervised way (i.e without labels). These methods generally involve a pretext task that is solved to learn a good representation and a loss function to learn with. Below you can find a continuously updating list of self-supervised methods.

Text 3, carried by 1 of 42 methods:

Attention Modules refer to modules that incorporate attention mechanisms. For example, multi-head attention is a module that incorporates multiple attention heads. Below you can find a continuously updating list of attention modules.

Text 4, carried by 1 of 42 methods:

Object Detection Models are architectures used to perform the task of object detection. Below you can find a continuously updating list of object detection models.

Text 5, carried by 1 of 42 methods:

Transformers are a type of neural network architecture that have several properties that make them effective for modeling data with long-range dependencies. They generally feature a combination of multi-headed attention mechanisms, residual connections, layer normalization, feedforward connections, and positional embeddings.

Text 6, carried by 1 of 42 methods:

Image Models are methods that build representations of images for downstream tasks such as classification and object detection. The most popular subcategory are convolutional neural networks. Below you can find a continuously updated list of image models.

Also reached at /methods/category/vision-transformer (Papers with Code's slug for this collection; the archive carries no slugs, so this site's is derived from the name).

Methods

All 42 methods in this collection, most-tagged first. Year is the archive's introduced_year; the archive stores 2000 when it has none, shown here as “–”. Papers counts distinct papers the archive tags with the method. Click a heading to sort.

Vision Transformer – 2,144
Swin Transformer – 416
DINO self-DIstillation with NO labels – 208
NesT – 37
Deformable DETR – 35
CCT Compact Convolutional Transformers – 33
PVT Pyramid Vision Transformer – 28
DPT Dense Prediction Transformer – 25
MobileViT – 22
CvT Convolutional Vision Transformer – 12
TNT Transformer in Transformer – 12
T2T-ViT Tokens-To-Token Vision Transformer – 11
MoCo v3 – 10
LV-ViT – 9
MViT Multiscale Vision Transformer – 9
CaiT Class-Attention in Image Transformers – 5
CoaT Co-Scale Conv-attentional Image Transformer – 5
MobileViTv2 – 5
Focal Transformers – 4
MUSIQ – 4
PVTv2 Pyramid Vision Transformer v2 – 4
XCiT – 4
ConViT – 3
CrossTransformers – 3
CrossViT – 3
VATT – 3
BatchFormer Batch Transformer – 2
CPVT Conditional Position Encoding Vision Transformer – 2
CeiT Convolution-enhanced image Transformer – 2
DeepViT – 2
LeVIT – 2
Colorization Transformer – 1
EsViT – 1
LocalViT – 1
MHMA Multi-Heads of Mixed Attention – 1
MobileViTv3 – 1
OODformer – 1
RegionViT – 1
Shuffle-T Shuffle Transformer – 1
Twins-PCPVT – 1
Twins-SVT – 1
Visformer – 1