add matcher: REDI-Match

[Vincentqyw]

📌 Repository

Repository URL: https://github.com/YinjiGe/REDI-Match.git

Paper (arXiv / conference): https://arxiv.org/abs/2606.24330

🔧 Matcher Info

Type:

• Sparse (extractor + matcher, e.g., SuperPoint+LightGlue)
• Dense / Standalone (end-to-end, e.g., LoFTR, RoMa)

Efficiency:

• High (real-time)
• Medium
• Low (slow, research-grade)

✅ Checklist

• The repository has an open-source license (MIT, Apache, BSD, etc.)
• The model has pretrained weights available
• The code is compatible with PyTorch
• I understand the matcher will be added as a git submodule under imcui/third_party/

📝 Notes