Meta announces ControlRoom3D Room Generation using Semantic Proxy Rooms paper page: huggingface.co/papers/2312.05… ControlRoom3D creates diverse and plausible 3D room meshes aligning well with user-defined room layouts and textual descriptions of the room style.

Thanks AK for sharing our work! 🙂

Our open-vocabulary OpenMask3D was featured in the RSIP Vision Computer Vision News!🗞️🚀 Thanks Ralph🙏 Read the full article by Ayça Takmaz and Elisabetta Fedele here: rsipvision.com/ComputerVision… --- Project: openmask3d.github.io🛋️ Paper: arxiv.org/abs/2306.13631📄 NeurIPS Conference

Hey everyone! We are thrilled to kick off the robot learning seminar and we are excited of having Francis Engelmann from ETH Zurich presenting tomorrow (Thursday 11:30am EST). See you soon! youtube.com/@MontrealRobot… #robotics #machinelearning #sceneunderstanding #foundationmodels

♻️Accelerating Diffusion Models through Block Caching♻️ During my internship Meta, we found that many layer computations during inference are redundant. With our caching method, we get 🏎 a speedup of 1.5x - 1.8x 🖼️ while maintaining similar quality! fwmb.github.io/blockcaching

Back for round two! Our workshop ☀️ OpenSUN 3D 🌍 on Open-Vocabulary 3D Scene Understanding is happening at #CVPR2024 in Seattle! Can't wait to see your creative paper submissions again! Deadline: April 1, 2024. Don't miss out! #OpenSUN3D

Check out our #CVPR'24 papers on 3D human interactions, generative 3D modeling, and uncertainty-aware and unsupervised 3D semantic scene understanding! Congrats to Lei Li David Rozenberszki Christian Diller Yawar Siddiqui Shivangi Jiapeng Tang Anh-Quan Cao for their amazing work!

New release of #PyViz3D 🚀-- directly render 3D point clouds, graphs and more in Blender 🔶 🤹🏻‍♂️ 👩‍💻Code+examples: github.com/francisengelma… 🛠️Install: pip3 install pyviz3d Good time to create beautiful figures for your European Conference on Computer Vision #ECCV2026 supplementary or #CVPR2026 workshop submissions✍️⏰

PyViz3D is a game changer - super easy to use, great for debugging with multi-layer support, and ideal for remote sharing. The icing on the cake: it now generates Blender files for publication-ready figures. Check it out! 🔥

Interested in the role of simulation in Autonomous Driving? Don't miss the opportunity to delve deeper into this fascinating topic with our stellar lineup of speakers at the #CVPR2024 workshop. Join us to explore cutting-edge insights and innovations!

📢Call for papers If you work on open-vocabulary 3D scene understanding, consider submitting your work to our #CVPR2024 workshop OpenSUN3D! ⌛️Deadline: April 1st, 2024🏃 Only 1 week left to submit your 8-page full papers or 4-page abstracts! More info: opensun3d.github.io

📢 Call for Papers We invite submissions to our #CVPR2024 workshop, OpenSUN3D, for open-vocabulary 3D scene understanding. We welcome both 4-page extended abstracts and 8-page submissions, including work that has already been published. ⌛️ Deadline: April 1st

The lack of 3D data is the key bottleneck in building foundation models for 3D. Our #CVPR2024 (Highlight) paper introduces a unified vision model that can utilize 2D pre-trained weights, web-scale data, & labelled 3D data for 2D & 3D perception tasks. odin-seg.github.io

I'm tenured at TU München! This wouldn't have been possible without the support of my incredible students and colleagues. Here's to more learning and discovery! Check out our 3DAI Lab: 3dunderstanding.org

A great success for GenAI Media Challenge Workshop, thanks for all the participants. All awards are given out and we had a great panel discussion about the future of GenAI. The benchmark and corresponding annotations are planned to be released. #cvpr

Our #CVPR2026 workshop on open-vocabulary 3D scene understanding is happening this afternoon in Seattle! ✨ Workshop schedule: opensun3d.github.io/#schedule We have a fantastic line-up of speakers 👇

🚗🔍 Announcing OoDIS: Anomaly Instance Segmentation Benchmark for Autonomous Vehicles! We extended popular anomaly segmentation datasets with instance and bounding box labels. 🦌📦 check it out at vision.rwth-aachen.de/oodis #CVPR2026 VAND 2.0 Workshop #CVPR2024

Join us at CVPR today for the PM session where we will be presenting ControlRoom3D! Location: Arch 4A-E, Poster #136 Time: 5pm We look forward to chatting with you about it! 😀

Excited to announce a research milestone that we have been working towards in GenAI 3D teams AI at Meta: Meta 3D Gen! End-to-end integration of fast, state-of-the-art text-to-3D generation and 3D texturing capabilities, for artist-created and generated 3D assets.