Probing the 3D Awareness of Visual Foundation Models

TL;DR

This paper investigates whether large-scale visual foundation models inherently encode 3D scene structure and surface consistency, revealing limitations through experiments with probes and zero-shot inference.

Contribution

It provides the first systematic analysis of 3D awareness in visual foundation models, highlighting their current limitations in representing 3D structure.

Findings

Models show limited 3D structural encoding.

Surface representation across views is inconsistent.

Current models lack robust 3D awareness.

Abstract

Recent advances in large-scale pretraining have yielded visual foundation models with strong capabilities. Not only can recent models generalize to arbitrary images for their training task, their intermediate representations are useful for other visual tasks such as detection and segmentation. Given that such models can classify, delineate, and localize objects in 2D, we ask whether they also represent their 3D structure? In this work, we analyze the 3D awareness of visual foundation models. We posit that 3D awareness implies that representations (1) encode the 3D structure of the scene and (2) consistently represent the surface across views. We conduct a series of experiments using task-specific probes and zero-shot inference procedures on frozen features. Our experiments reveal several limitations of the current models. Our code and analysis can be found at https://github.com/mbanani/probe3d.