ShapeSplat: A Large-scale Dataset of Gaussian Splats and Their Self-Supervised Pretraining Paper: arxiv.org/abs/2408.10906 Project: unique1i.github.io/ShapeSplat Video: youtu.be/5TnfkO3WdMk?si… Method ⬇️ 1 I 2

Wisdom

Check out Yongshuo (Yongshuo Zong), Tim (Timothy Hospedales), and my survey paper on Self-Supervised Multimodal Learning. Updated version now on arxiv.

TrackGo A Flexible and Efficient Method for Controllable Video Generation discuss: huggingface.co/papers/2408.11… Recent years have seen substantial progress in diffusion-based controllable video generation. However, achieving precise control in complex scenarios, including

Results from GLOMAP -> 3DGS. I'd say it worked really well! Total COLMAP time including undistortion and image matching was around 2 minutes. 3DGS model training took 16 minutes to 30k steps.

Katie Link This one should be a lot better for clinical scenarios than existing models. "Towards Evaluating and Building Versatile Large Language Models for Medicine." arxiv.org/abs/2408.12547. Github: github.com/MAGIC-AI4Med/M…… Leaderboard: henrychur.github.io/MedS-Bench/

I'm sorry if I scared you . Countless Buddha heads have been stolen from China. They are in British Museum，the Guilmet Museum in Paris，Metropolitan Museum of Art and so on.

Oh wow! I just tested Splatt3R with my own data on my computer, which creates 3D Gaussian Splats at 4 FPS on uncalibrated 512x512 2D images! It's by far the fastest 3D reconstruction method, powered by MASt3R. Check out the video!

Please enjoy the online huggingface.co/spaces/brandon… by Brandon Smart. A simple pipeline yet achieves realistic 3D reconstruction in the wild with uncalibrated images.

Wait! You think you need at least two images to reconstruct a scene? Think again! With GenWarp, you can reconstruct a 3D scene from just a single image! Simply create two views from a single input image and drop them into Splatt3R. What a time to be alive! 😉

Italy is boosting young researchers' careers!✨🚀 The 🇮🇹 Ministero dell'Università e della Ricerca will fund 250 former #MSCA postdoctoral fellows or Seal of Excellence holders with €67.5M. 📅 Deadlines: • Postdoc: 1/10 • Seal of Excellence holders: 19/12 🔗Apply now! ➡️ europa.eu/!rMM8py

Diffusion Policy Policy Optimization paper page: huggingface.co/papers/2409.00… We introduce Diffusion Policy Policy Optimization, DPPO, an algorithmic framework including best practices for fine-tuning diffusion-based policies (e.g. Diffusion Policy) in continuous control and robot

Are you a science fan? Join us for European Researchers' Night on 27.09!🌍🔬 Enjoy hands-on experiments, science shows, games, and more—all FREE & open to everyone! Bring your family, friends, or come solo. Check out the map for events near you! More➡️ europa.eu/!JdQRBv

Want to extract detailed meshes from 3DGS without without slow TSDF fusion? In Gaussian Opacity Fields (GOF) we show that it's possible to directly identify the geometry's level set and turn it into a compact and adaptive mesh! Background included. niujinshuchong.github.io/gaussian-opaci…

I bet generating human-object interactions will soon play a bigger role, as it not only helps with populating virtual environments, but also potentially serves as a coarse target for humanoid robots (e.g. language → HOI → control) 🧵

How can we generate high-fidelity, complex 3D scenes? Quan Meng's LT3SD decomposes 3D scenes into latent tree representations, with diffusion on the latent trees enabling seamless infinite 3D scene synthesis! w/ Lei Li, Matthias Niessner quan-meng.github.io/projects/lt3sd/

Today, we're announcing Pixtral 12B! It's our first vision model, and is really strong on the standard multimodal benchmarks, **without compromising abilities on your favourite text and reasoning tasks**. Blog: mistral.ai/news/pixtral-1… Le Chat 🐈: chat.mistral.ai/chat

Want to work on cutting-edge research in #AI? We have several 𝐏𝐡𝐃 & 𝐏𝐨𝐬𝐭𝐃𝐨𝐜 𝐨𝐩𝐞𝐧𝐢𝐧𝐠𝐬 in our Visual Computing & AI Lab in Munich! Topics have a strong focus on GenAI, including 3DGs, NeRFs, Diffusion Models, LLMs, etc. Share & apply :) application.vc.in.tum.de