Philippe Weinzaepfel
@weinzaepfelp.bsky.social
Principal Research Scientist in Computer Vision at Naver Labs Europe https://philippeweinzaepfel.github.io/
We just released our #ECCV2026 paper on Model Merging for Computer Vision 🎓 arxiv.org/abs/2604.12935 Joint work w @pdejorge.bsky.social Cesar De Souza @bjoernmichele.bsky.social @mbsariyildiz.bsky.social @weinzaepfelp.bsky.social Florent Perronnin & @skamalas.bsky.social See Pau's thread below 🔽
1/6 Excited to share that our paper on model merging was accepted at ECCV 2026! 🎉 We introduce an efficient, decoder-free proxy that makes model selection faster, simpler and practical across vision tasks. 📄 arxiv.org/abs/2604.12935 🌐 europe.naverlabs.com/task-alignment 🧵👇
#ECCV2026 paper: A scalar per patch from pre-trained ViTs enables fast moving navigation in the real world 966 *REAL* nav episodes (!!) performed by S. Janny with Dino-v3, Dino-v2, DUNE, VC1, AM-RADIO encoders show that patch features can be bottlenecked to 1 value ➡️ affordances emerge. 1/8
This is work at Naver Labs Europe by - Philippe Weinzaepfel (@weinzaepfelp.bsky.social) - Yours truly, - Mert Bulent Sariyildiz (@mbsariyildiz.bsky.social) - Guillaume Bono, - Gianluca Monaci arxiv.org/abs/2606.21562 7/8
For your Embodied AI task you want a recurrent model with constant complexity per step, but you don't want to lose the power of transformers (which store the full obs history and attend to it)? Do not despair, we have your back. We distill transformers into recurrent transformers 1/8
Interested in designing the next generation of FF3D reconstruction models for real-life use cases? The Geometric Deep Learning team at the root of the foundational research lines of CroCo and DUSt3R is looking for a research scientist to help us!
Multi-HMR 2 just dropped changing real-time human-centered scene understanding💣! 1 forward pass delivers: 🤼 multi-person detection 🫂 3D body mesh recovery for all ages (🙏 Anny) 📐 camera params+metric-scale positions 👣 identity tracking 📍Paper+code+blog+interactive demo 🎮🚀 tinyurl.com/multi-hmr-2
Join our #cvpr tutorial, where we will unpack human body models — from SMPL to Anny — and their use in RGB and LiDAR-based Mesh Recovery. With István Sárándi, Salma Galaaoui, @fbaradel.bsky.social, Nermin Samet and @davidpicard.eurosky.social. Program: human-mesh-tutorial.github.io
Could your HMR prediction do with a bit of refinement? Try Anny-Fit – a new framework for all-ages that improves existing models & also generates pseudo GT for in the wild images🔥! 📄Anny-Fit Code & Paper: tinyurl.com/anny-fit #CVPR2026 #3DVision Anny (Apache 2.0): tinyurl.com/anny-model
In my team we have two open positions on AI for robotics and in particular on manipulation, for junior and senior researchers. With us, connect fundamental research questions with real operational problems and contribute to emerging AI-driven robotics services. careers.werecruit.io/en/naver-lab...
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
How to scale visual perception in robotics w/o scaling compute? We present a distilled 'universal encoder' that's compact and compatible with multiple 2D&3D vision tasks! → up to 90% less encoding memory 🗜️ → 4× faster at inference 🚀 More info -->> tinyurl.com/universal-en...
Cycling break after #ECCV2026 deadline (well at day+1 as the deadline was at 11pm in France)
Philippe Weinzaepfel@weinzaepfelp.bsky.social · 10mo agoA bit of cycling after the #CVPR2026 deadline rush
I have been asked several times how we draw the 3D Figures for of our papers, so I wrote a blog post on it. This does not replace a Blender tutorial, just a couple specifics for scientific papers: - import images as planes - render edges - transparent BG chriswolfvision.medium.com/creating-3d-...
Check out our 2025 highlights in computer vision! 🚀Five new *St3R models (MASt3R-SfM, MUSt3R, PanSt3R, HAMSt3R, HOSt3R) 🤩Anny parametric 3D human model (Apache 2.0) 🤟Universal encoder for all-in-one vision FM Watch the highlights 👇 More info ▶️ tinyurl.com/muvs5vnu
I will present "Kinaema" this FRI at 11am at NeurIPS 2025. Drop by if you are interested in sequence models for robotics (recurrent transformers), map-free pose estimation from memory and navigation! europe.naverlabs.com/research/pub... B. Sariyildiz, P. Weinzaepfel, G. Bono, G. Monaci and myself.
The 4th #AI4RoboticsWorkshop is over! A big THANK YOU to all our fab speakers & participants for great presentations & conversations. Snapshot souvenir ⬇️ For those who missed out -recordings available soon! tinyurl.com/bdtk2nzs
We’re ready to roll for Day 2 of the #AI4RoboticsWorkshop ! Livestream starts 9am CET:🎥 tinyurl.com/bdtk2nzs Today’s great lineup of speakers: @dcremers.bsky.social – @ericbrachmann.bsky.social - Aniruddha Kembhavi - Adrien Gaidon - Nicolas Mansard & Justin Carpentier
We’re live! 🚀 Streaming: tinyurl.com/bdtk2nzs The International Workshop on AI4Robotics by @naverlabseurope 2dys of Spatial AI, SLAM, robot learning, HRI, autonomy This AM CET: @martinhumenberger.bsky.social @marcpollefeys.bsky.social Andrea Vedaldi Cordelia Schmid & @andrewdavidson.bsky.social ⬇️
Martin Humenberger kicks off Naver Labs Europe's 4th AI for Robotics Workshop in Meylan, France, talking about DUSt3R, MASt3R, MUSt3R etc. europe.naverlabs.com/updates/ai4r... @naverlabseurope.bsky.social @martinhumenberger.bsky.social
Marc Pollefeys talks about Global SfM meeting feedforward reconstruction, targeting scenes with a high number of hops between cameras to cover the entire scene (and many other Spatial AI contributions) @marcpollefeys.bsky.social AI 4 Robotics Workshop at @naverlabseurope.bsky.social
🧍♀️ Introducing Anny: an open, interpretable, and differentiable human body model for all ages. Grounded in anthropometric data (MakeHuman) & WHO stats, Anny offers: 🧠 Interpretable shape control 👶👩🦳 Unified from infants to elders 🧩 Versatile for fitting, synthesis & HMR 🌍 Open under Apache 2.0
Romain Brégier@rbregier.bsky.social · 10mo agoMeet Anny, our Free (Apache 2.0) and Interpretable Human Body Model for all ages. Anny is built upon #MakeHuman and enables achieving SOTA performance in Human Mesh Recovery. ArXiv: arxiv.org/abs/2511.03589 Demo: anny-demo.europe.naverlabs.com Code: github.com/naver/anny
Meet Anny. One model. Every body. A new human model that fits everyone! ✅ Works for all ages ✅ Free & open (Apache 2.0) ✅ Privacy-friendly (no scans) ✅ Simple parameters Blog: tinyurl.com/5fsekm9z Code: github.com/naver/anny An alternative 4 #VR, #AR #Robotics. #3DHumanModeling #OpenSource
Meet Anny, our Free (Apache 2.0) and Interpretable Human Body Model for all ages. Anny is built upon #MakeHuman and enables achieving SOTA performance in Human Mesh Recovery. ArXiv: arxiv.org/abs/2511.03589 Demo: anny-demo.europe.naverlabs.com Code: github.com/naver/anny
"Sliding is all you need" (aka "What really matters in image goal navigation") has been accepted to 3DV 2026 (@3dvconf.bsky.social) as an Oral presentation! By Gianluca Monaci, @weinzaepfelp.bsky.social and myself. @naverlabseurope.bsky.social
Christian Wolf@chriswolfvision.bsky.social · last yr.In a new paper led by Gianluca Monaci, with @weinzaepfelp.bsky.social and myself, we explore the relationship between rel pose estimation and image goal navigation and study different architectures: late fusion, channel cat (w/ or w/o space2depth) and cross-attention. arxiv.org/abs/2507.01667 🧵1/5
We have a new sequence model for robotics, which will be presented at #NeurIPS2025: Kinaema: A recurrent sequence model for memory and pose in motion arxiv.org/abs/2510.20261 By @mbsariyildiz.bsky.social, @weinzaepfelp.bsky.social, G. Bono, G. Monaci and myself @naverlabseurope.bsky.social 1/9
A reminder which might be relevant now: we are looking to hire a senior research scientist in Robotics at @naverlabseurope.bsky.social in Grenoble, France.
