Project AURIONAI Use-Cases Research & Innovation
Microsoft AI
visionfamily

Swin Transformer

Microsoft hierarchical vision transformer with shifted windows; a general backbone for classification, detection and segmentation.

Microsoft AIMicrosoft AIUSA
Figure from the Swin Transformer paper: the shifted window approach for computing self-attention.
Levelfamily
Classvision
Parametersnot in the notes
Contextnot in the notes
Released25 Mar 2021
Opennessunknown
Runs onmobile or NPU, CPU server
Statusactive

Lineage

20212022Swin-BSwin-B, Mar 2021Swin-T, Mar 2021Swin Transformer V2Swin Transformer V2, Nov 2021
Microsoft AISwin Transformer V2 belongs to this familyvision18 Nov 2021
Microsoft AISwin-B belongs to this familyvision25 Mar 202187.83M
Microsoft AISwin-T belongs to this familyvision25 Mar 2021
Where this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. Confidence and evidence are the notes' own.

R&D Topics

Image classification backbone architectures (CNN and vision transformer)exemplar

Identity

Developer
Microsoft AI
Family
not in the notes
Released
25 Mar 2021 sourcesannouncement page
Weights published
–
Record checked
12 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.

Architecture and Size

Architecture
dense transformer
Total parameters
– sources
Active parameters
–
Checkpoint parameters
–
Note
Swin T, S, B, L; Swin V2 up to 3B parameters; Video Swin is a separate record.
Context length
not in the notes
Input
image
Output
not in the notes
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).

Openness and Access

Openness
unknown
Licence
not in the notes sources
Access routes
open weights download, cloud marketplace
Where this comes from
The licence file governs over registry tags and aggregator labels.

Deployment

Runs on
mobile or NPU, CPU server
Footprint
Edge or toolkit builds listed in: Qualcomm AI Hub, OpenVINO.
Runtimes
not in the notes
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.

Indonesia

Indonesian support
not stated
Regional languages
not in the notes
Jakarta hosting
unknown
Languages
not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.

Benchmark Results

The notes record no benchmark result for this model.

Industry Use Cases

No industry use case links this model.

Sources

  1. arxiv.org/abs/2103.14030 https://arxiv.org/abs/2103.14030
  2. jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json https://jumpstart-cache-prod-us-west-2.s3.us-west-2.amazonaws.com/models_manifest.json
  3. huggingface.co/microsoft/swin-base-patch4-window7-224 https://huggingface.co/microsoft/swin-base-patch4-window7-224
  4. ai.azure.com/catalog/models?task=image-classification https://ai.azure.com/catalog/models?task=image-classification
  5. huggingface.co/microsoft/swinv2-tiny-patch4-window16-256 https://huggingface.co/microsoft/swinv2-tiny-patch4-window16-256
  6. aihub.qualcomm.com/models/swin_base https://aihub.qualcomm.com/models/swin_base
  7. github.com/pytorch/vision/blob/main/torchvision/models/s... https://github.com/pytorch/vision/blob/main/torchvision/models/swin_transformer.py

To cite this page: Project AURION, AI Use-Cases Research & Innovation, "Swin Transformer", https://aurion.semai.id/models/microsoft-swin-transformer/, records of October 2026.