David Nordström
@davnords.bsky.social
Phd Student @ Chalmers Deep Learning for Computer Vision. Strengthen your ViTs: https://github.com/davnords/octic-vits
New matcher unlocked. Also some VGGT-interp stuff
Zhenjun Zhao@ericzzj.bsky.social · 2mo agoRoMa-Ω: What Feed-Forward 3D Models Know About Image Matching @davnords.bsky.social, Xinyue Zhang, @thibautloiseau.bsky.social, @vincentlepetit.bsky.social, @fredkahl.bsky.social tl;dr: replace the DINOv3 in RoMa v2 with features from the VGGT-Ω arxiv.org/abs/2609.09507
Heading to #CVPR26 now! You'll find me at the Image Matching Worskhop on June 4th talking about rotation invariant matching and on June 6th presenting MuM as a poster. If you are in Denver, let's meet up!
Image matching since 2020: 2020: @pesarlin.bsky.social SuperGlue 2023: @vincentleroy.bsky.social DUSt3R 2024: @parskatt.bsky.social RoMa 2025: @jianyuanwang.bsky.social VGGT 2026: @davnords.bsky.social"hold my beer" scales LightGlue) 2026: @jianyuanwang.bsky.social "no, hold MY beer"(scales VGGT)
It seems, that we have failed the communication about IMC26. Let's try again. The competition this year is here: kaggle.com/competitions... No prizes, but whole year leaderboard -- similar to KITTY and other academic competitions. 3D people, please retweet and share.
Image Matching Challenge 2025 Ongoing
Ongoing leaderboard for Image Matching Challenge 2025.
kaggle.com
LoMa-R is our newest addition to the LoMa family. In our IMW paper at #CVPR26, we investigate rotation invariance in the sparse matching pipeline. The resulting model is robust to rotations, even matching star constellations, and achieves strong upright performance. github.com/davnords/loma
GitHub - davnords/LoMa: LoMa: Local Feature Matching Revisited
LoMa: Local Feature Matching Revisited. Contribute to davnords/LoMa development by creating an account on GitHub.
github.com
Johan Edstedt@parskatt.bsky.social · 7mo agoAccepted to the Image Matching Workshop at #CVPR26!
Sparse image matching is done via 1) keypoint detection in each image, 2) keypoint description, 3) matching of descriptions between images. Should rotation invariance be enforced at stage 2 or 3? Turns out both work fine! To be presented at the CVPR image matching workshop by @davnords.bsky.social
Johan Edstedt@parskatt.bsky.social · 7mo agoIntroducing LoMa-R! A rotation invariant version of LoMa. Code: github.com/davnords/loma Paper: arxiv.org/abs/2604.11809
Introducing LoMa-R! A rotation invariant version of LoMa. Code: github.com/davnords/loma Paper: arxiv.org/abs/2604.11809
Johan Edstedt@parskatt.bsky.social · 7mo agoIntroducing LoMa, the next generation of feature matcher!
LoMa: Local Feature Matching Revisited @davnords.bsky.social @parskatt.bsky.social , @bokmangeorg.bsky.social et 6 al tl;dr: if you train DeDoDe+LightGlue on VGGT-scale data, it helps a LOT. New IMC2022 sota arxiv.org/abs/2604.04931
Introducing LoMa, the next generation of feature matcher!
Yesterday, @parskatt.bsky.social and the RoMa v2 team got an award at the annual Swedish Symposium on Image Analysis. FeelsGoodMan. @bokmangeorg.bsky.social @fredkahl.bsky.social @jastermark.bsky.social @liucvl.bsky.social et. al.
Biggest psyop: Learning rate decay. @parskatt.bsky.social showed me the truth but I refused to accept it...
Presenting today at #BMVC2025 our follow-up work on anisotropic rotation averaging which is particularly useful in global SfM. We propose a fast solver ACD and integrate robust optimization. If you’re at the conference, welcome to come to our poster #516! bmvc2025.bmva.org/proceedings/...
We are introducing MuM, a feature encoder (ViT-L) tailored for 3D vision tasks. TLDR; Spiritual successor to CroCo with a simpler multi-view objective and larger scale. Beats DINOv3 and CroCo v2 in RoMa, feedforward reconstruction, and rel. pose. arxiv.org/abs/2511.17309 github.com/davnords/mum
Johan out here carrying Swedish CV academia with a 10 author absolute masterclass. Only the avatar can master all four elements (LiU, CTH, LTH and UvA). Very strong and versatile model overall. RoMa v2 can match anything you'd like. Even @bokmangeorg.bsky.social 's hand annotated satellite images.
Johan Edstedt@parskatt.bsky.social · 11mo agoRoMa v2 is now out! (github.com/Parskatt/rom..., arxiv.org/abs/2511.15706) Here are the main improvements we made since RoMa:
RoMa v2: Harder Better Faster Denser Feature Matching @parskatt.bsky.social et 11 al. tl;dr: in title. Predict covariance per-pixel, more datasets, use DINOv3, adjust architecture. arxiv.org/abs/2511.15706
Turns out NLP is just vision
mr. TIM@timkellogg.me · last yr.Z.ai released a paper very similar to DeepSeek-OCR on the same exact day (a few hours earlier afaict) Glyph is just a framework, not a model, but they got Qwen3-8B (128k context) to handle over 1 million context by rendering input as images arxiv.org/abs/2510.17800
These gentlemen show how not only colmap but also VGGT fail on spherical motion in their ICCV oral paper "Uncalibrated Structure from Motion on a Sphere". I wonder if it is just a data issue for VGGT or if it is deeper than that. I mean VGGT was trained on mostly synthetic data. What do you think?
Pro tip: For good Halloween vibes, use non-normalized RoPE on images larger than your training resolution and larger than the composite period of some of the RoPE-rotations. You might get scary ghost structures in your features.
Towards the Next Generation of 3D Reconstruction @parskatt.bsky.social PhD Thesis. tl;dr: would be useful in teaching image matching - nice explanations. (too) Fancy and stylish notation. Cool Ack section and cover image. liu.diva-portal.org/smash/record...
And here is a link to the thesis itself: liu.diva-portal.org/smash/record...
Towards the Next Generation of 3D Reconstruction
liu.diva-portal.org
How to name your method: a comprehensive flow chart
Born too late to explore the earth. Born too early to explore the galaxy. Born just in time to \nabla_{\theta}f_{\theta}
