Evaluating Robustness of Monocular Depth Estimation with Procedural Scene Perturbations

Jack Nugent · Siyang Wu · Zeyu Ma · Beining Han · Meenal Parakh
Abhishek Joshi · Lingjie Mei · Alexander Raistrick · Xinyuan Li · Jia Deng

Princeton University

Paper

We evaluate the robustness of monocular depth estimation models to controlled 3D scene perturbations including object, camera, material, and lighting changes. This repository includes the dataset and evaluation code for measuring model robustness.

Procedural Depth Evaluation (PDE) Dataset

The dataset can be downloaded here.

Citation

If you find our Procedural Depth Evaluation useful, please consider citing our academic paper:

@misc{nugent2025evaluatingrobustnessmonoculardepth,
      title={Evaluating Robustness of Monocular Depth Estimation with Procedural Scene Perturbations}, 
      author={Jack Nugent and Siyang Wu and Zeyu Ma and Beining Han and Meenal Parakh and Abhishek Joshi and Lingjie Mei and Alexander Raistrick and Xinyuan Li and Jia Deng},
      year={2025},
      eprint={2507.00981},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2507.00981}, 
}