🔥Telegram İfşa

Tired of SLAM breaking in dynamic scenes? WildGS-SLAM (#CVPR2025) uses uncertainty-aware tracking & DINOv2-based maps to handle motion, remove dynamic objects, and enable high-quality view synthesis — all from monocular RGB. 👇 🌐wildgs-slam.github.io 📜arxiv.org/abs/2504.03886

The use of AI in reviewing is a growing problem. Several of my ICCV papers have AI reviews -- one reviewer was so lazy that they left in the prompts! A common refrain that I hear is that people have difficulty writing in English and need to use AI to clean up their review.

Excited to present our #CVPR2025 paper DepthSplat next week! DepthSplat is a feed-forward model that achieves high-quality Gaussian reconstruction and view synthesis in just 0.6 seconds. Looking forward to great conversations at the conference!

SupeRANSAC: One RANSAC to Rule Them All Baráth Dániel tl;dr: why RANSAC work well for different vision problems? ->answer->implementation details and problem-specific optimizations arxiv.org/abs/2506.04803

SupeRANSAC: One RANSAC to Rule Them All Baráth Dániel tl;dr: most accurate, but slow-ish RANSAC, by incorporating best practices & tricks: arxiv.org/abs/2506.04803 github.com/danini/superan… Sampling: PROSAC + P-NAPSAC for homography & rigid pose, PROSAC for epipolar 1/

For those going to #ICCV2025 - welcome to our RANSAC tutorial on October 2025 with - Baráth Dániel - Viktor Larsson - Eric Brachmann - Jiri Matas - and me. danini.github.io/ransac-2025-tu… #ICCV2025

ReSplat: Learning Recurrent Gaussian Splats Haofei Xu, Baráth Dániel, Andreas Geiger, Marc Pollefeys tl;dr: rendering error->recurrent network->Gaussian updates arxiv.org/abs/2510.08575

🚀Excited to share our recent work on test-time scaling for feed-forward Gaussian splatting: we learn a recurrent model ReSplat that is able to iteratively improve the reconstruction quality in a feed-forward manner! haofeixu.github.io/resplat/

🙏Big thanks to Baráth Dániel, Andreas Geiger, and Marc Pollefeys for a great collaboration on this project!

HouseTour: A Virtual Real Estate A(I)gent Ata Celen, Marc Pollefeys Baráth Dániel Iro Armeni tl;dr: 3D trajectory for walkthrough + text summary from 3D reconstruction. openaccess.thecvf.com/content/ICCV20…

Planar Affine Rectification from Local Change of Scale and Orientation Yuval Nissan, Marc Pollefeys Baráth Dániel tl;dr: in title. P.S. "Planar Affine Rectification from Change of Scale" was in 2010 %) openaccess.thecvf.com/content/ICCV20…

We are ready for the second day of #ICCV2025 🌺 you can find us at: 🐊 CroCoDL Workshop: 301B, 1:00PM ➡️localizoo.com/workshop ✨RANSAC Tutorial: 328, 1:00PM ➡️danini.github.io/ransac-2025-tu… Zuri Bauer Hartwig Hermann Blum Baráth Dániel Marc Pollefeys #ICCV2025

That's where you want to be this afternoon at #ICCV2025 in room 328. Dmytro Mishkin 🇺🇦 Baráth Dániel Eric Brachmann

#ICCV2025 Dmytro Mishkin 🇺🇦 Baráth Dániel Eric Brachmann Open Problems & Conclusions of the tutorial Open: -RANSAC properly for dense (like 20k+ correspondences)? -auto inlier-outlier threshold selection? -integrate in end-to-end recon methods. -differentiable RANSAC that has all components to be SOTA? -RANSAC failure detection

If you’re attending ICCV, come see our papers! - HouseTour: A Virtual Real Estate A(I)gent, Wed, 2:45 pm - Planar Affine Rectification from Local Change of Scale and Orientation (highlight), Thu, 2:45 pm - Hierarchical 3D Scene Graphs Construction Outdoors, Thu, 2:45 pm

Visit our Demo on Controllable 3D Object Generation #ICCV2025 - today afternoon Oct Tue 21, 3pm-5pm - Exhibit Hall 1 Come early, so we can catch some waves before sunset 🌊🏄🏝️ Project: spacecontrol3d.github.io Marc Pollefeys Elisabetta Fedele Ian Huang Or Litany Leonidas Guibas

Come check this afternoon: 🏡 HouseTour: A Virtual Real Estate A(I)gent Posed photos ➡️ 3D tours + summaries. Built on 1.6K+ real house tours with poses, reconstructions & descriptions.   📍#ICCV2025 | Poster #275 🔗 house-tour.github.io W/ Ata Celen, Baráth Dániel, Marc Pollefeys

YoNoSplat: You Only Need One Model for Feedforward 3D Gaussian Splatting Botao Ye, Boqi Chen, Haofei Xu, Baráth Dániel, Marc Pollefeys tl;dr: first use ground-truth poses then mix of predicted and ground-truth poses for training; intrinsics->scale ambiguity arxiv.org/abs/2511.07321