1. Motivation

• In this paper, it proposed the method to combine vision transformer architecture and self-supervied learning

---

2. Method

• Overall structure
  

2.1. Self-supervised learning

• 

2.2. SwinUNet

• This work proposed two separate SwinUNet streams for contrastive learning process as Fig 2.

---

3. Result

---

Reference

• https://openaccess.thecvf.com/content/CVPR2022W/EarthVision/papers/Scheibenreif_Self-Supervised_Vision_Transformers_for_Land-Cover_Segmentation_and_Classification_CVPRW_2022_paper.pdf