#CO3Dv2 dataset and challenge are now live!

3D computer vision research just got easier! We’re releasing Implicitron, an extension of PyTorch3D that enables fast prototyping of 3D reconstruction and new-view synthesis methods based on rendering of implicit representations.

⏳ Time is running out, share your CO3D challenge submission by October 19. #ECCV2022

We are hiring PhD student interns to work on 3D/4D reconstruction / conditional generation! Feel free to get in touch if you want to apply or get more information. metacareers.com/jobs/721145616…

Consider presenting your accepted CVPR'23 work on 3D CV at this CVPR workshop and participate in the 3D object segmentation and material estimation challenge! Deadline for challenge submissions June 1st: t.ly/BJWzh.

My first PhD paper with Visual Geometry Group (VGG) and @MetaAI, on depth estimation from dynamic stereo videos has been accepted at CVPR 2023!! ☺️ #CVPR2023 #cvpr #computervision #3D

CoTracker: It is Better to Track Together paper page: huggingface.co/papers/2307.07… Methods for video motion prediction either estimate jointly the instantaneous motion of all points in a given video frame using optical flow or independently track the motion of individual points

New on Hugging Face — CoTracker simultaneously tracks the movement of multiple points in videos using a flexible design based on a transformer network — it models correlation of the points in time via specialized attention layers. 🤗 Try CoTracker ➡️ bit.ly/3swQFqt

Thrilled to share Mosaic-SDF (M-SDF), a simple 3D representation suitable for 3D generative models! Check out more results of text-to-3D generations here - lioryariv.github.io/msdf

1/5 Last week, we released an updated version of CoTracker, our transformer-based model for tracking any point in a video. It’s now capable of tracking 10x more points than before!

Meta presents IM-3D Iterative Multiview Diffusion and Reconstruction for High-Quality 3D Generation paper page: huggingface.co/papers/2402.08… Most text-to-3D generators build upon off-the-shelf text-to-image models trained on billions of images. They use variants of Score

Excited to release Lightplane: A highly memory-efficient differentiable NeRF renderer and feature splatter. It renders FullHD image batches using <1GB of GPU memory and greatly scales up 3D deep models such as the Large Reconstruction Model. With Ang Cao, Justin Johnson, Andrea

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.

New AI research from Meta – CoTracker3 Simpler and Better Point Tracking by Pseudo-Labelling Real Videos. More details ➡️ go.fb.me/xiyc63 Demo on Hugging Face ➡️ go.fb.me/yzuqd0  Building on our previous work on CoTracker, this new model demonstrates impressive

We are releasing uCO3D! Built to supercharge 3D GenAI and digital-twin models, this evolution of CO3D features more and higher-quality object videos from 1k categories, 3D Gaussian Splats, and streamlined OSS tools. 💻Data&code: github.com/facebookresear… 📄Paper:

New dataset from researchers at Meta — uCO3D, or UnCommon Objects in 3D, is the largest publicly-available object-centric dataset for 3D deep learning and 3D generative AI. More on this project ➡️ go.fb.me/8u86hq

Natalia Neverova You absolutely cooked on this release. Very proud ♥️

BREAKING: Meta's Llama 4 Maverick just hit #2 overall - becoming the 4th org to break 1400+ on Arena!🔥 Highlights: - #1 open model, surpassing DeepSeek - Tied #1 in Hard Prompts, Coding, Math, Creative Writing - Huge leap over Llama 3 405B: 1268 → 1417 - #5 under style control