PhD. Student @ Chalmers Computer Vision and Deep Learning Code: https://teapawt.net/t.co/R54G3HJNlo
Joined February 2022
- Tweets384
- Following193
- Followers281
- Likes882
Pinned Tweet
Today we release Poincar3, our newest 3D model trained entirely through SSL on RGB image sequences.
I find the coolest thing is that, without labels, it learns multi-view geometry. Just using the attention map, reliable tracks can be formed (see below).
davnords.com/poincar3
David Nordström retweeted
To match or not to match?
FlexGlue 🐙, our new image-matching model, can match every pixel AND reject doppelganger pairs, tackling a common failure case in SfM.
➡️State-of-the-art SfM and matching, blazing fast
➡️Code: github.com/cvg/FlexGlue
➡️Paper: arxiv.org/pdf/2610.32976
1/
I would be remiss if I did not mention that the naming and motivation originates from the SeeSE(3) paper (now accepted to NeurIPS 2026) by @maks_ovsjanikov and others. arxiv.org/abs/2607.14228
1
92
ICLR template's font and vibe is so damm cool. While single-column is annoying, the paper just instantly looks nice
279
Today at #ECCV2026 we will present our work. At 9AM we will present LoMa and RoMa v2 on stage in AB.
You will find us in the 10.30AM poster session with LoMa (#106), RoMa v2 (#99), and Quick ViTs (#26).
David Nordström retweeted
RoMa-Ω: What Feed-Forward 3D Models Know About Image Matching
@davnords, Xinyue Zhang, @thibaut_loiseau, @VincentLepetit2, @FredrikKahl
tl;dr: replace the DINOv3 in RoMa v2 with features from the VGGT-Ω
arxiv.org/abs/2609.09507
David Nordström retweeted
It is highly suspicious that RoMa v2 author @davnords wears T-shirt “I ❤️ ROMA” and not “I ❤️ ROMA v2”
