François Rameau

my fav papers from 3DV because why not 🤝 MapAnything by Meta → 1B foundation model for 3D → takes in images + optional ray directions, poses & depth outputs 3D, normal, depth maps → consists of mm encoders on top of transformer with MLP, DPT and pose head model & demo ➡️

The Best Paper Award of #3DV2026 goes to "Look Around and Pay Attention: Multi-camera Point Tracking Reimagined with Transformers"! Congratulations!

Everyone's excited about Karpathy's autoresearch that automates the experiment loop. We automated the whole damn thing. 🦞 Meet AutoResearchClaw: one message in, full conference paper out. Real experiments. Real citations. Real code. No human in the loop. One message in → full paper out. Here's what happens in between: 📚 Raids arXiv & Semantic Scholar, digests 50+ papers in minutes 🥊 Three AI agents FIGHT over the best hypothesis (one swings big, one sanity-checks, one tries to kill every idea) 💻 Writes experiment code from scratch, adapts to your hardware 💥 Code crashes at 3am? It reads the stack trace, rewrites the fix, keeps going 🔄 Results weak? It pivots to entirely new hypotheses and starts over 📝 Drafts a full paper with citations, every single one verified against live databases No babysitting. No Slack messages. No "hey can you re-run this." Karpathy built the experiment loop. We built the whole lab. Chat an idea. Get a paper. 🦞 Try it 👉: github.com/aiming-lab/Aut… Kudos to the team @JiaqiLiu835914, @richardxp888, @lillianwei423, @StephenQS0710, @Xinyu2ML, @HaoqinT, @zhengop, @cihangxie, @dingmyu, and we are looking for more contributors.

⁦@mbzuai⁩ winter ML school

@kwangmoo_yi At least you got a reply from the reviewer ;-)

Begin rant If something is shown to work, and the authors explain it, but you don't understand, it's not nice to stick to a suspicion that you have that is empirically already disproved. End rant

Reminder: the #ECCV2026 submission portal is now open.

The day has arrived, somewhere in the world 😅 Our Program Chairs and senior Area Chairs are going through the final checks. Only the Program Chairs know the exact timing of the release. The rest of us will find out at the same time. Good luck!

@ducha_aiki @CSProfKGD All the cases I have witnessed this year had a decent rebuttal. One of my papers is affected, and a friend of mine has three papers with at least one missing decision in his batch. From what I see online, many authors are facing the same problem.

@Frameau89 @CSProfKGD Are you excluding the case when there was no rebuttal submitted?

And that’s a wrap for my #CVPR2026 AC duties

@CSProfKGD Thanks a lot for your kind reply. It is so much extra work for ACs to deal with those cases. Hopefully, not too many papers were affected!

I’ve seen similar reports on social media. Would be interesting to hear from PCs, if this was indeed widespread. My experience was some reviewers were late with final recommendations but did submit it eventually. Some were still in discussions with the other reviewers at the (soft) discussion deadline. Yes, ACs should consider the reviews but also verify that the concerns raised were addressed in the rebuttal. These reviews still contain valuable signals. No consequences as far as I’m aware of but reviewers do get rated by ACs and to me this is grounds for a low rating and comment that they did not fulfill their full reviewer duties.

Pixel-Accurate Epipolar Guided Matching Oleksii Nasypanyi, Francois Rameau tl;dr: if you do (for some reason) epipolar guided matching, this is probably the optimal way? Definitely improves upon our epipolar hashing from 2021 openreview.net/forum?id=9zRX5… #3DV2026

@ducha_aiki Thanks for sharing our work! It is super niche, but we had a lot of fun working on it. Epipolar hashing was definitely a strong source of inspiration.

A Global AI Learning Festival in South Asia 11 days of learning, curiosity, and collaboration came to a close but the impact continues far beyond the final ceremony. Until we meet again Lectures videos: youtube.com/watch?v=eoMb6A… See you at the 7th edition of ANAIS later in 2026.

We’re excited to announce the release of our Annual Newsletter 2024, Issue 3, where you’ll discover how NAAMII is driving breakthroughs across research, education, and industry collaborations. Click the link below: naamii.org.np/newsletters/ Happy reading!

Meet @Frameau89 , Assistant Professor at SUNY in Korea, who was a speaker at ANAIS 2023. He will be joining us again this year at 5th ANAIS 2024, so don't miss your chance to learn from this incredible speaker, along with 14+ other speakers from around the world!

ANAIS 2024 – The Global AI Experience You've Been Waiting For! 🚀 Register Now!! lnkd.in/gcDFQ92Y 📅 Second Round Application Deadline: 20th October, 2024 🌐 Open to National and International Participants For more information visit: lnkd.in/d7jBe7Gb

MC-Calib v1.3.0 comes with improvements to the utilities for post-calibration analysis as well as bug fixes in non-overlapping camera calibration scenario github.com/rameau-fr/MC-C…

HUGE shoutout to our #CVPR2024 Outstanding Reviewers 🫡

Cannot overestimate the power of the open-source - a user discovered a bug that caused worsened calibration of the fisheye setup. Now, MC-Calib v1.2.1 fixes it and brings improved calibration of fisheye cameras github.com/rameau-fr/MC-C…

@AntoineGrondin Around me, people tend to prefer Conda for its simplicity. I do not see virtual-env widely used these days. But, personally, I believe that if you are already familiar with Docker, it is the best option. It offers better repeatability, and Docker Hub is very convenient ;-)

Python package management/runtime dependencies: how to do it well? pip + virtual env? nix? pip + docker? something else? Purpose is to play with ML stuff, scientific computing, so GPU.