REMM:Rotation-Equivariant Framework for End-to-End Multimodal Image Matching

The objective of this research is Rotation-Equivariant Framework for End-to-End Multimodal Image Matching. It will be a long time for the release of full codes since the paper is under review. Pretrained weights will be able soon...

result

(a)OPT-SAR：SEN2 & SEN1

(b)OPT-SAR：Google Earth & GF-3

(c)OPT-NIR: Aerial image & Aerial image

(d)OPT-NIR: Camera & Camera

Preprint

https://arxiv.org/abs/2407.11637