Post

USat A Unified Self-Supervised Encoder for Multi-Sensor Satellite Imagery

Posted
By 1 min read
  • 论文名称: USat: A Unified Self-Supervised Encoder for Multi-Sensor Satellite Imagery
  • Link: https://arxiv.org/abs/2312.02199
  • Published in: Arxiv 2023
  • Type: Pretrain
  • Code/Project: https://github.com/stanfordmlgroup/USat
  • 备注: 新的USat backbone，多光谱，MAE预训练
  • 数据类型: 多光谱
  • Backbone: Transformer
  • 下游任务: Classification, Semantic Segmentation
  • Short Summary: 遥感数据包含多传感器、多光谱和时间信息，提供了大量可用于自监督预训练的自标注数据,开发了一种新的编码器架构，称为 USat，可以输入来自多个传感器的多光谱数据进行自监督预训练
Arxiv
paper Pretrain
This post is licensed under CC BY 4.0 by the author.