Astra-vision
Computer vision group at Astra, Inria Paris
Oct. 2023
We study 2D vision and 3D perception for robust scene understanding. Our research focuses on relaxing the use of abundant data and supervision, stepping towards weak-/un-supervised vision algorithms, while providing models that are more interpretable. Our work addresses general scene understanding, with a focus on mobile robotics.
Research axes
Learning with less supervision
Vision in complex conditions
3D scene understanding
news
| Oct 24, 2026 | We’re opening a PhD on “Multimodal physically-consistent world modeling”, (⚠️ 2 deadlines: October 31st and November 20th). Details in the job page |
|---|---|
| Oct 23, 2026 | KoalaGS is accepted as spotlight @NeurIPS 2026 (0.9% acceptance rate) 🎉 |
| Oct 1, 2026 | One in, one out. Mohammad Fahes just defended his great PhD (kudos) and Shivam Tripathi is freshly starting a PhD. |
| Sep 25, 2026 | Welcoming Denis Akola from NYU for a day visit, and a talk on Novel Depth Synthesis with VFM. |
| Apr 1, 2026 | Happy to welcome Seamie Hayes for a PhD visit. |
| Nov 1, 2025 | Welcoming new members in the group: Fatima Baldé and Antonios Tragoudaras starting a PhD, Radu Beche for a PhD visit and Sebastian Cavada as visitor and collaborator. |
| Jul 13, 2025 | Organizing the 2nd African Computer Vision Summer School in Kigali, Rwanda from July 13th to 23rd, with an amazing lineup of speakers. |
| Jun 27, 2025 | MatSwap was awarded 🏆 Best Paper Honorable Mention at EGSR 25. Kudos to Ivan and team. |
| Jun 12, 2025 | Organizing the PixFoundation workshop on Pixel-level Vision Foundation Models at CVPR 2025, in Nashville, USA. |
| Jun 2, 2025 | Happy to host Fabio Pizzati, postdoc at MBZUAI, for a one month visit. Welcome Fabio 👋😃 |
| Apr 25, 2025 | 6 group members received a CVPR 25 Outstanding Reviewer Award. With 6 outstanding reviewers (bravo to Alexandre, Gilles, Ivan, Quan, Mohammad, Yasser) and 3 Area Chairs (Andrei, Raoul, Renaud), Astra-Vision is making its mark to CVPR. |
| Apr 15, 2025 | LiDPM is accepted as oral (approx. 7%) at IV 2025. Kudos to Tetiana. |
| Apr 15, 2025 | 🎓 We opened a PhD topic on “Physics-Grounded Vision Foundation Models”. Apply before May 20th 2025. More details on the job page. |
| Apr 15, 2025 | Our paper FLOSS is out with code. It builds on the intruiging observations that some CLIP templates excel at given semantic segmentation classes. FLOSS formulates a new training-free Open-vocabulary Semantic Segmentation that can be plugged into any methods. Bravo Yasser and team. |
| Feb 1, 2025 | Welcoming two new interns in our group: Fatima Balde and Jonathan Seele. Welcome and good luck for your work. |
| Oct 20, 2024 | New papers building on VLMs: 📎 ProLIP shows that adequately finetuning the last CLIP layer significantly boost few-shot classification. ☕ LatteCLIP proposes an unsupervised prototype-based CLIP finetuning from only synthetic labels. |
| Oct 1, 2024 | The group organized the first ACVSS summer school in Kenya and the 3rd WSCV workshop in Deep Learning Indaba in Dakar. |
| Jul 1, 2024 | 🧠 UMBRAE: Unified Brain Decoding is accepted to ECCV 2024. |
| May 25, 2024 | 🕵️ PaSCo selected as Best Paper Award Candidate of CVPR 2024. |
| Feb 26, 2024 | Our 3 papers got accepted to CVPR 2024: 🎨 Material Palette, 🍴 FAMiX, 🕵️ PaSCo. |