Pi3 EU AI Act Compliance Profile

🌌 $\pi^3$: Scalable Permutation-Equivariant Visual Geometry Learning

<div align="center"> <p> <a href="https://huggingface.co/papers/2507.13347" target="_blank"> <img src="https://img.shields.io/badge/Paper-00AEEF?style=plastic&logo=arxiv&logoColor=white" alt="Paper"> </a> <a href="https://yyfz.github.io/pi3/" target="_blank"> <img src="https://img.shields.io/badge/Project Page-F78100?style=plastic&logo=google-chrome&logoColor=white" alt="Project Page"> </a> <a href="https://github.com/yyfz/Pi3" target="_blank"> <img src="https://img.shields.io/badge/GitHub-Code-blue?logo=github" alt="GitHub"> </a> <a href="https://huggingface.co/spaces/yyfz233/Pi3" target="_blank"> <img src="https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Demo-blue" alt="Hugging Face Demo"> </a> </p> </div> <div align="center"> <a href="https://yyfz.github.io/pi3/"> <img src="https://huggingface.co/yyfz233/Pi3/resolve/main/main.png" width="90%"> </a> <p> <i>$\\pi^3$ reconstructs visual geometry without a fixed reference view, achieving robust, state-of-the-art performance.</i> </p> </div>

✨ Overview

We introduce $\pi^3$ (Pi-Cubed), a novel feed-forward neural network that revolutionizes visual geometry reconstruction by eliminating the need for a fixed reference view. Traditional methods, which rely on a designated reference frame, are often prone to instability and failure if the reference is suboptimal.

In contrast, $\pi^3$ employs a fully permutation-equivariant architecture. This allows it to directly predict affine-invariant camera poses and scale-invariant local point maps from an unordered set of images, breaking free from the constraints of a reference frame. This design makes our model inherently robust to input ordering and highly scalable.

A key emergent property of our simple, bias-free design is the learning of

yyfz233's obligations as provider

These are yyfz233's responsibility, not yours.

Provider obligations are being compiled.

📋

No personalized actions yet