Evaluating Robustness of Monocular Depth Estimation with Procedural Scene Perturbations

TL;DR

This paper introduces PDE, a benchmark for systematically evaluating the robustness of monocular depth estimation models against various scene perturbations, revealing their vulnerabilities and guiding future improvements.

Contribution

The paper presents PDE, a novel procedural benchmark for robustness testing of monocular depth estimation, addressing limitations of standard accuracy-focused evaluations.

Findings

Certain perturbations significantly degrade model performance.

Robustness varies across different scene changes.

Insights inform future model development for better resilience.

Abstract

Recent years have witnessed substantial progress on monocular depth estimation, particularly as measured by the success of large models on standard benchmarks. However, performance on standard benchmarks does not offer a complete assessment, because most evaluate accuracy but not robustness. In this work, we introduce PDE (Procedural Depth Evaluation), a new benchmark which enables systematic robustness evaluation. PDE uses procedural generation to create 3D scenes that test robustness to various controlled perturbations, including object, camera, material and lighting changes. Our analysis yields interesting findings on what perturbations are challenging for state-of-the-art depth models, which we hope will inform further research. Code and data are available at https://github.com/princeton-vl/proc-depth-eval.