🚀 Introducing InstantSfM: Fully Sparse and Parallel Structure-from-Motion. ✅ Python + GPU-optimized implementation, no C++ anymore! ✅ 40× faster than COLMAP with 5K images on single GPU! ✅ Scales beyond 100 images (more than VGGT/VGGSfM can consume)! ✅ Support metric scale.

Our recent research will be presented at #ICCV2025 #ICCV2025! We’ll present 5 papers about: 💡 self-supervised & representation learning 🌍 3D occupancy & multi-sensor perception 🧩 open-vocabulary segmentation 🧠 multimodal LLMs & explainability valeoai.github.io/posts/iccv-2025

Éloi Zablocki Alexandre Boulch Mickael Chen Matthieu Cord Sophia Sirko-Galouchenko Spyros Gidaris Antonín Vobecký Andrei Bursuc nicolasthome FLOSS: Free Lunch in Open-vocabulary Semantic Segmentation by: Yasser Benigmim, M. Fahes, Tuan-Hung VU, Andrei Bursuc, R. de Charette. 📄: arxiv.org/abs/2504.10487 x.com/yasserbenigmim…

Today Corentin Dancette from Raidium stopped by to talk about Curia, a multi-modal open-weights foundation model for radiology building in the style of DINOv2 with tons of downstream applications. Do check it out!

Working with CLIP for classification, segmentation, or detection? Want to boost performance without labels and while staying training-free? Come see our poster #157 (Session 1) tomorrow!! we’ll tell you more about FLOSS: Free Lunch in Open-Vocabulary Semantic Segmentation!

Come to the FLOSS poster #157 on Thursday #ICCV2025 to find out more about this really simple strategy for boosting open-vocabulary semantic segmentation and meet Yasser Benigmim & Tuan-Hung VU

Thrilled to have Jayneel Parekh visiting us over a break in the CVPR sprint to talk about his exciting works in analyzing MLLMs representations. Check out his great work jayneelparekh.github.io

Nano Banana Pro: "Generate a diagram of a two-layer neural network in the style of Stephen Biesty"

Honored to host a very special guest today: Huy V. Vo, PhD alumnus of valeo.ai. He presented his work on automatic data-curation strategies for self-supervised representation learning (DINOv2, DINOv3). Find out more about his research here: huyvvo.github.io

[RA-L 2025] ActiveSplat: High-Fidelity Scene Reconstruction through Active Gaussian Splatting github.com/Li-Yuetao/Acti… ActiveSplat enables the agent to explore the environment autonomously to build a 3D map on the fly. The integration of a Gaussian map and a Voronoi graph assures

We have a few fun papers at #BMVC2025 Check them out!

Need pixel-level features from your backbone (DINOv3, CLIP, RADIO, FRANCA...)? 🚀Introducing NAF: A universal, zero-shot feature upsampler. It turns low-res ViT features into pixel-perfect maps. -⚡ Model-agnostic -🥇 SoTA results -🚀 4× faster than SoTA -📈 Scales up to 2K res.

Off-the-shelf feature upscaling from any vision model without retraining. 🥇 SoTA results ⚡️ High throughput 🧠 Zero-shot generalization 😍 Code available 👇 Check out the details!

Check out NAF: an effective ViT feature upsampler to produce excellent (and eye-candy) pixel-level feature maps. NAF outperform both VFM-specific upsamplers (FeatUp, JAFAR) and VFM-agnostic methods (JBU, AnyUp) over multiple downstream tasks 👇

🎉 Try NAF live — now on Hugging Face! 🤗 🖼️ Upload any image 🤖 Select any VFM (DINOv2, CLIP, RADIO, FRANCA...) 📐 Choose any resolution ⚡ Instant upsampling Try it now: huggingface.co/spaces/LChambo…

1/Serve your PEFT with a fresh IPA!🍺 Finetuning large models is cheaper thanks to LoRA, but is its random init optimal?🤔 Meet IPA: a feature-aware alternative to random projections #NeurIPS2025 WS #CCFM Oral+Best Paper Work w/Shashank Tuan-Hung VU Andrei Bursuc Matthieu Cord 🧵

Check out our works at NeurIPS Conference #NeurIPS2025 this week! We present 5 full papers + 1 workshop about: 💡 self-supervised & representation learning 🖼️ generative image models 🧠 finetuning & understanding LLMs & multimodal LLMs 🔎 feature upsampling valeoai.github.io/posts/neurips-…

I'm speaking at #aiPULSE2025 today on Open & re-purposable foundation models for the automotive industry. The morning keynotes talked a lot about open source so my slide here might be timely.

Quick clarification: we first shared the Outstanding Paper Award thinking it was the top prize — and it turns out IPA did win Best Paper! Absolutely thrilled! 🎉

Today in our TUM AI - Lecture Series we'll have the amazing Ruiqi Gao, Google DeepMind. She'll talk about "𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐠𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐯𝐞 𝐰𝐨𝐫𝐥𝐝 𝐦𝐨𝐝𝐞𝐥𝐬: progress and challenges". Live stream: youtube.com/live/CkOSMqwvF… 7pm GMT+1 / 10am PST (Tue Dec 16th).