Contributing meaningfully to NeurIPS-level research this early is no small achievement. It has been a pleasure to watch Emirhan and Sarthak bring curiosity, persistence, and fresh thinking to our group. Congratulations to both! 👏 #MachineLearning #YoungResearchers #NeurIPS

Feed-forward models distort geometry even when given camera poses. In paper to appear in #NeurIPS, we recast multi-view stereo as sequence-to-sequence: a camera-aware transformer with a unified global cost volume predicts geometry for all views jointly. arxiv.org/abs/2609.24850

Excited that our paper DUET is accepted by @NeurIPSConf! Check it out: github.com/Haoyu-Hu/DUET Thanks for the great help from my wonderful collaborators @norijacoby @Orson_Xu @xuandongzhao More details coming soon! #NeurIPS #RL #ReinforcementLearning

Compositional SDF models often produce objects that interpenetrate. Soft penalties reduce this without eliminating it. In a paper to appear at #NeurIPS, we enforce hard non-intersection constraints plug-and-play and with negligible overhead. arxiv.org/pdf/2605.29761 #3DVision

Excited to share our new work, 𝗩𝗶𝗧𝗲𝗫-𝗕𝗲𝗻𝗰𝗵: 𝗕𝗲𝗻𝗰𝗵𝗺𝗮𝗿𝗸𝗶𝗻𝗴 𝗛𝗶𝗴𝗵-𝗙𝗶𝗱𝗲𝗹𝗶𝘁𝘆 𝗩𝗶𝗱𝗲𝗼 𝗦𝗰𝗲𝗻𝗲 𝗧𝗲𝘅𝘁 𝗘𝗱𝗶𝘁𝗶𝗻𝗴, accepted to the #NeurIPS 2026 𝗘&𝗗 𝗧𝗿𝗮𝗰𝗸! 🎉 Editing text in videos is surprisingly challenging: a good model needs to generate the correct text, keep it temporally consistent across frames, and preserve everything else in the scene. To study this problem better, we introduce 𝗩𝗶𝗧𝗲𝗫-𝗕𝗲𝗻𝗰𝗵, a comprehensive benchmark for video scene text editing. Our release includes: 🔹 𝗩𝗶𝗧𝗲𝗫-𝗗𝗮𝘁𝗮𝘀𝗲𝘁 — 387 real-world 720p videos with text-region masks and editing instructions 🔹 𝗔 𝘁𝗵𝗿𝗲𝗲-𝗮𝘅𝗶𝘀 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝗽𝗿𝗼𝘁𝗼𝗰𝗼𝗹 covering text correctness, temporal quality, and edit locality with 13 metrics 🔹 𝗘𝘅𝘁𝗲𝗻𝘀𝗶𝘃𝗲 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝗼𝗳 𝟴 𝗯𝗮𝘀𝗲𝗹𝗶𝗻𝗲𝘀 across four video-editing paradigms 🔹 𝗩𝗶𝗧𝗲𝗫-𝗘𝗱𝗶𝘁-𝟭𝟰𝗕, an open-source reference model with motion-aligned glyph-video conditioning 🔹 Open-source 𝗱𝗮𝘁𝗮, 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 𝗰𝗼𝗱𝗲, 𝗺𝗼𝗱𝗲𝗹 𝘄𝗲𝗶𝗴𝗵𝘁𝘀, 𝗮𝗻𝗱 𝗹𝗲𝗮𝗱𝗲𝗿𝗯𝗼𝗮𝗿𝗱 One takeaway from our study: no single metric tells the whole story. Models that preserve the scene well may fail to actually edit the text, while models that generate the right text frame-by-frame can still suffer from severe temporal flickering. ViTeX-Bench is designed to make these trade-offs explicit and reproducible. We hope ViTeX-Bench can provide a useful foundation for future research on precise and temporally consistent video editing. 🔗 Project page: vitex-bench.github.io 📜 Paper link: arxiv.org/abs/2609.40356 #NeurIPS2026 #ComputerVision #GenerativeAI #VideoEditing #VideoGeneration #MultimodalAI #MachineLearning #AIResearch

🎉 Proud to share that two of my works have been accepted to #NeurIPS2026 — AdapToPASS as a Spotlight 🌟 and SphMind as a Poster 📌 (main) ! Excited to present these works at #NeurIPS and connect with the community!

#NeurIPS results are out! Whether your paper is accepted or rejected, you still have a chance to present at InfPriv @ #NeurIPS2026, Sydney! Consider submitting your work via our NeurIPS fast-track with minimal effort, where the deadline is 9/25 AOE. beyond-private-training.ai.s…

I am thrilled to announce that our paper, "Differentiating Network Design Objectives for Balancing Cost and Distance," has been accepted to #NeurIPS 2026! Congratulations to Zishang on his first top-conference paper at RUC! 🎉

👏 We’ve got some extra #NeurIPS News just now! Our Position Track Paper was also distinguished for an Oral presentation at the Conference! 🎯🎯 Read more here: linkedin.com/feed/update/urn… @MunichCenterML @zuseschoolrelAI @eyke_hu @ysale12