ROVR-Open-Dataset: A Large-Scale Depth Dataset for Autonomous Driving

TL;DR

ROVR is a large-scale, diverse, and cost-effective depth dataset for autonomous driving, covering various scenarios and conditions, with tools for easy reproduction and robust model training.

Contribution

The paper introduces ROVR, a scalable, multi-region depth dataset with accompanying tools, addressing limitations of existing datasets in diversity, cost, and scalability.

Findings

Extensive ablation studies on scene types, illumination, weather, and sparsity.

Identification of three failure modes in current architectures: photometric collapse, geometric confusion, and range saturation.

Public release of dataset, tools, and evaluation code.

Abstract

Depth estimation is a fundamental component of spatial perception for autonomous driving and other unmanned systems operating in open urban environments. Existing depth datasets such as KITTI, nuScenes, and DDAD have advanced the field but are limited in diversity and scalability, and benchmark performance on them is approaching saturation. A less discussed constraint is \emph{sensor economics}: the bespoke multi-LiDAR rigs behind these datasets are expensive, power-hungry, and difficult to replicate at fleet scale, which caps the geographic and temporal diversity that any single benchmark can cover. We present ROVR, a large-scale, diverse, and cost-efficient depth dataset designed to capture the complexity of real-world driving. ROVR comprises 200K high-resolution frames across highway, rural, and urban scenarios, spanning day/night cycles and adverse weather conditions, collected across North America, Europe, and Asia. We additionally release the calibration, synchronization, preprocessing, and privacy pipeline so that the platform can be reproduced by third parties. The lightweight acquisition pipeline enables scalable collection, while sparse but statistically sufficient ground truth -- validated by a density ablation -- supports robust model training. Extensive ablation studies further characterize performance across scene types, illumination, weather conditions, and ground-truth sparsity levels, and identify three qualitatively distinct failure modes -- photometric collapse, geometric confusion, and range saturation -- that current architectures share. The dataset, data loaders, calibration and privacy pipelines, and evaluation code are publicly available at \url{https://xiandaguo.net/ROVR-Open-Dataset}.