TartanMatch adapts UFM, an RGB-pretrained correspondence model, to support five modalities within one shared network.
One model matches pixels across any pair of five sensor modalities
TartanMatch unifies RGB, event, thermal, depth, and projected LiDAR in a single network, cutting endpoint error on zero-shot benchmarks by up to 61.9%.
Top University
Hyeokjoon Kwon · Jiting Cai · Ruogu Li · Kritan Bhandari · Geethika Hemkumar · Parv Maheshwari · +4 more
Carnegie Mellon University · Fisk University
Research Digest··2 min read
The authors present TartanMatch, a single network that predicts dense pixel correspondence for any ordered pair among RGB, event, thermal, depth, and projected LiDAR.
Why this paper
From Carnegie Mellon University and Fisk University
In one line
A single model achieves state-of-the-art dense correspondence across all 25 pairs among five sensor modalities.
What we could check
- ·No code link found
- ·No weights link found
- ·No dataset link found
- ·No compute details found
- ✓Limitations stated by the authors
- ·No benchmark numbers found
Observed from the paper text and links we have. Absence here means we did not find it, not that it does not exist.
§