@visinf

Visual Inference Lab of @stefanroth at @TUDarmstadt. Research in Computer Vision and Machine Learning.

 Darmstadt, Germany
visinf.tu-darmstadt.de
Joined April 2012
  • Tweets192
  • Following395
  • Followers869
59Media
Visual Inference Lab retweeted
ELLIS@ELLISforEuropeOct 1
🚨 Apply now for the #ELLISPhD Program! Work with leading AI/ML researchers across Europe, with exchange visits & joint supervision. The application deadline is 31 Oct. Get all info & apply now at ellis.eu/news/ellis-phd-prog…
18
4
77
12,712
Visual Inference Lab retweeted
Georgia Chalvatzaki
@GeorgiaChalSep 19
I wrote a blog sharing some thoughts after spending the last days looking more carefully at the GPT-6 Astra results in robotics, 3D, CAD and autonomous experimentation. I started rather skeptical, and I still don’t think these results mean that scaling has suddenly “solved robotics.” But I do think something interesting is happening. What capability are we actually seeing? What role does structure play? What is learned by the model, and what becomes possible because of the tools, representations and algorithms around it? And perhaps most importantly, what experiments do we actually need to answer these questions? What Are We Actually Seeing with GPT-6 Astra? #Robotics #GPT6 #Astra #RobotLearning xitter.cf/GeorgiaChal/status/210…
Georgia Chalvatzaki
@GeorgiaChalSep 19

What Are We Actually Seeing with GPT-6 Astra?

Article · @GeorgiaChal
9
103
17,664
1/6 🧵 ReconSplat: Generalizable 3D Scene Reconstruction Beyond Observed Views by @GpStracquadanio, Kevin Raj, @JuliaGrabinski, @stefanroth 📍 Thursday Poster Session 1 @ECCV Poster #196
1
7
40
3,043
more replies
5/6 ☁️ From novel views to 3D reconstructions. We can unproject our predictions into point clouds, revealing sharp and consistent geometry, even *beyond the observed views*. ✨ Jointly generating appearance and geometry lets us complete scenes in 3D.
1
1
147
6/6 🔗 Links to our work. Paper: arxiv.org/abs/2608.28895 Website: visinf.github.io/reconsplat Code: github.com/visinf/reconsplat Video: youtube.com/watch?v=ht-JDxqB…

ReconSplat: Generalizable 3D Scene Reconstruction Beyond Observed Views

We introduce ReconSplat, a feed-forward model for 3D scene reconstruction that aims to address the longstanding trade-off between plausible view generation for unobserved regions and geometric...

arxiv.org
2
115
Visual Inference Lab retweeted
Kosta Derpanis
@CSProfKGDSep 8
Dinner with the keynote speakers and organizers of the #ECCV2026 Workshop on Explainable Computer Vision. Thank you @sukrutrao @AdaGorgun @schaub_simone Stephan Alaniz @Amin_Prc for the invitation! Great hanging out with @LilyWeng_ @HildeKuehne my fellow speakers too!
4
4
62
5,237
[1/6] 📢 We are in Malmö at #ECCV2026 presenting 5 papers! 🎉
1
7
1
28
4,398
more replies
[5/6] LeAD-M3D: Leveraging Asymmetric Distillation for Real-Time Monocular 3D Detection J. Meier*, J. Michel*, O. Dhaouadi, @YungHsuYang, @ChristophReich1996, @zuriich92, @stefanroth, @mapo1, J. Kaiser, D. Cremers (@tumcvg) 🌍 Project Page: deepscenario.github.io/LeAD-…
1
3
133
[6/6] MAGneT-3D: Monocular and Domain-Generalizable Temporal 3D Detection M. Kotb, J. Meier, @ChristophReich1996, O. Dhaouadi, L. Denninger, D. Cremers (@tumcvg) 📚 Paper: arxiv.org/abs/2608.14282

MAGneT-3D: Monocular and Domain-Generalizable Temporal 3D Detection

Monocular temporal 3D detection aims to detect objects in 3D, given a monocular video. Query-based 3D detectors unify detection and cross-view association, but their learnable queries fit the...

arxiv.org
1
4
66
Visual Inference Lab retweeted
Claudia Cuttano@ClaudiaCuttanoJun 15
Had so much fun at #CVPR2026! 🎉 Really happy to have presented two papers this year and discussed them with so many people. Both projects came out of my time at @visinf, TU Darmstadt, last year. Very grateful to @stefanroth for hosting me and helping shape them from the beginning, and to all my co-authors for being part of this 😊
4
4
94
5,824
Visual Inference Lab retweeted
SkalskiP
@skalskip92Jun 7
INSID3 segments objects across domains using ONLY ONE annotated example it works entirely without a segmentation decoder, task-specific fine-tuning, or external mask generators like SAM CVPR 2026 paper with enormous practical potential
8
80
4
625
55,622
Visual Inference Lab retweeted
Gabriele Trivigno@gabTrivvJun 6
Today! I'm going to present MARCO in Oral Session 4D at 2 pm, and later at poster #20! Don't miss it :)
Claudia Cuttano@ClaudiaCuttanoMay 25
✨#CVPR2026 Oral ✨ A tale of a failed experiment: what if you fine-tune DINOv2 on sparse keypoints, beat every benchmark, only to discover it performs worse than the original frozen model on novel keypoints? 🚀MARCO closes this gap: a unified model for generalisable correspondences github.com/visinf/MARCO
2
8
2
52
9,866
Visual Inference Lab retweeted
Claudia Cuttano@ClaudiaCuttanoJun 6
✨ #CVPR2026 Oral ✨ INSID3 turns a frozen DINOv3 into a training-free in-context segmenter across domains and granularities! Excited to present our work today in Oral Session 4D (14:00–15:15). Come by our poster later if you’d like to chat: 📍 Poster #19 🕓 16:00–18:00 See you there!
4
19
3
126
10,542
📢 [CVPR’26] Can we learn to detect, segment, and track every object in a video without human supervision?  Yes, we introduce VideoCUPS, the first unsupervised video panoptic segmentation (VPS) method: 1. Get pseudo-labels from monocular videos. 2. Train a VPS model on them. – at Denver, CO
8
72
3
425
27,923
more replies
When fine-tuned with just 10% of labels, VideoCUPS already matches a fully supervised model trained on all Cityscapes-VPS labels, and outperforms the DINO-initialized baseline significantly.
1
6
674
Work by: @ChristophR1996*, @olvr_hhn*, @neekans, @lealtaixe, C. Rupprecht, D. Cremers and @stefanroth 📄Paper: arxiv.org/abs/2606.04925 🌍Project Page: visinf.github.io/videocups/ 💻Code: github.com/visinf/cups 📹Video: youtube.com/watch?v=lDvgajKB… 👁️CVPR: Friday, Poster Session 2 #333
3
15
695
Load more