Skip to content

Latest commit

 

History

History
74 lines (51 loc) · 4.61 KB

related_file.md

File metadata and controls

74 lines (51 loc) · 4.61 KB

Related Resources for Swin Transformer

This document provides a curated list of useful resources for understanding and utilizing the Swin Transformer, including papers, video tutorials, and talks.

📄 Papers

  1. Swin Transformer: Hierarchical Vision Transformer using Shifted Windows

    • Authors: Ze Liu, Yutong Lin, Yue Cao, Han Hu, et al.
    • Published: 2021
    • Summary: The foundational paper introducing Swin Transformer, discussing the shifted window mechanism, hierarchical structure, and the model’s applications to object detection and semantic segmentation.

  2. Swin Transformer V2: Scaling Up Capacity and Resolution

    • Authors: Ze Liu, Han Hu, Yutong Lin, et al.
    • Published: 2021
    • Summary: An extension to the original Swin Transformer paper, SwinV2 scales up the model for higher resolutions and increased capacity, achieving state-of-the-art results on COCO and ADE20K datasets.

  3. Visual Transformers: Token-based Image Representation and Processing

    • Authors: Kaiming He, Ross B. Girshick, et al.
    • Published: 2020
    • Summary: This paper explores the use of transformers in computer vision, serving as a background for Swin Transformer’s architectural choices.

  4. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale (ViT)

    • Authors: Alexey Dosovitskiy et al.
    • Published: 2020
    • Summary: The Vision Transformer (ViT) paper that initiated the trend of using transformers in computer vision. Swin Transformer builds on some of the limitations of ViT, such as computational complexity with high-resolution images.

🎥 Video Tutorials

  1. Swin Transformer Explained - by Yannic Kilcher

    • Channel: Yannic Kilcher
    • Duration: 20 minutes
    • Summary: A concise explanation of Swin Transformer’s architecture, key components, and why it’s more efficient for vision tasks compared to other models like ViT. Great for a high-level overview.

  2. Swin Transformer for Vision Tasks - Official Microsoft Research Presentation

    • Channel: Microsoft Research
    • Duration: 45 minutes
    • Summary: In-depth talk by one of the authors explaining Swin Transformer’s development, challenges, and solutions, with a focus on real-world applications.

  3. Transformers for Computer Vision - Introduction and Applications

    • Channel: Two Minute Papers
    • Duration: 10 minutes
    • Summary: An introduction to how transformers are used in computer vision, with Swin Transformer as one of the examples. Ideal for beginners.

📢 Talks & Presentations

  1. CVPR 2021 Workshop - Swin Transformer: Scaling Vision Transformers

    • Event: CVPR 2021
    • Description: This workshop includes discussions on the Swin Transformer model, its scalability, and performance benchmarks across various vision tasks.

  2. Microsoft Research Webinar - Advances in Vision Transformers: From ViT to Swin Transformer

    • Event: Microsoft Research Webinar Series
    • Description: A webinar by Microsoft Research focusing on the development of vision transformers, covering the progression from ViT to Swin Transformer and its use in production applications.

  3. NeurIPS 2021 Tutorial on Vision Transformers

    • Event: NeurIPS 2021
    • Description: A tutorial session that includes Swin Transformer among other vision transformers, exploring both theoretical and practical aspects of transformer models in vision tasks.

🗂️ Other Resources

These resources will help you understand Swin Transformer in depth, including its architectural innovations, applications, and advantages in computer vision tasks.