Introducing VGGT (CVPR'25), a feedforward Transformer that directly infers all key 3D attributes from one, a few, or hundreds of images, in seconds! No expensive optimization needed, yet delivers SOTA results for: ✅ Camera Pose Estimation ✅ Multi-view Depth Estimation ✅ Dense Point Cloud Reconstruction ✅ Point Tracking Project... Page: Code & Weights:show more

Try our demo live on Hugging Face Spaces! 🤗: (See demo illustration below) 👇

The key idea is designing a simple architecture with minimal 3D inductive biases to learn from ample quantities of 3D-annotated data. In this work, we demonstrate the benefit of training a single network to learn multiple 3D quantities simultaneously, even when these outputs may potentially overlap. Additionally, we propose Alternating-Attention, which enables the transformer to focus alternately within each frame and globally.

A strong advantage of our method is the ability to predict 3D attributes without any expensive optimization. For example, 🔸 VGGT can easily process ~200 images in ~10s on a single 40GB A100 GPU 🔸 50x faster than usual optimization-based methods, using far less memory.

Bonus insight: Using pretrained VGGT significantly enhances downstream tasks like: 🚀 Non-rigid point tracking 🚀 Feed-forward novel view synthesis

Compared to concurrent CVPR'25 Transformer-based 3D reconstruction methods, VGGT achieves significantly higher accuracy, with speed similar to the fastest variant Fast3R.

⚡Interesting observation⚡: VGGT’s camera & depth predictions are highly accurate and consistent. Unprojecting our predicted depth with predicted camera parameters yields even more precise point clouds than directly predicted point maps! Try this yourself using the Hugging Face demo 🤗

Project Page: Code & Weights: Great work together with @MinghaoChen23 , @n_karaev , Andrea Vedaldi, Christian Rupprecht, @davnov134 ! @Oxford_VGG @AIatMeta

Tried out on our my very cluttered robot room with my phone camera, works amazingly well! Nice work

I tried it on custom data, its works amazingly. So far the best results I have got.

Looks great