MultiCorrupt

<p align="center"> <img src="docs/multicorrupt_transparent.png" align="center" width="75%"> <h3 align="center"><strong>MultiCorrupt: A Multi-Modal Robustness Dataset and Benchmark of LiDAR-Camera Fusion for 3D Object Detection</strong></h3> <p align="center"> <a href="https://www.linkedin.com/in/tillbeemelmanns/" target='_blank'>Till Beemelmanns</a>¹&nbsp;&nbsp; <a href="https://google.com" target='_blank'>Quan Zhang</a>²&nbsp;&nbsp; <a href="https://www.linkedin.com/in/christiangeller/" target='_blank'>Christian Geller</a>¹&nbsp;&nbsp; <a href="https://www.ika.rwth-aachen.de/de/institut/team/univ-prof-dr-ing-lutz-eckstein.html" target='_blank'>Lutz Eckstein</a>¹&nbsp;&nbsp; <br> <small>¹Institute for Automotive Engineering, RWTH Aachen University, Germany&nbsp;&nbsp;</small> <br> <small>²Department of Electrical Engineering and Computer Science, TU Berlin, Germany&nbsp;&nbsp;</small> </p> </p>

Abstract: Multi-modal 3D object detection models for autonomous driving have demonstrated exceptional performance on computer vision benchmarks like nuScenes. However, their reliance on densely sampled LiDAR point clouds and meticulously calibrated sensor arrays poses challenges for real-world applications. Issues such as sensor misalignment, miscalibration, and disparate sampling frequencies lead to spatial and temporal misalignment in data from LiDAR and cameras. Additionally, the integrity of LiDAR and camera data is often compromised by adverse environmental conditions such as inclement weather, leading to occlusions and noise interference. To address this challenge, we introduce MultiCorrupt, a comprehensive benchmark designed to evaluate the robustness of multi-modal 3D object detectors against ten distinct types of corruptions.

arXiv | IEEE Explore | Poster | Trailer | Dataset Download

<!-- omit in toc -->

Overview

• Corruption Types
• News
• Benchmark Results
• Metrics
• Dataset Download
• Dataset Compilation
• Dataset Usage
• TODOs
• Contribution
• Acknowledgments
• Citation

Corruption Types

Missing Camera

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |--------------------------------------------------------|--------------------------------------------------------|--------------------------------------------------------| | Multi-View: | Multi-View: | Multi-View: | | | | |

Motion Blur

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Front: | Front: | Front: | | | | |

Points Reducing

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Front: | Front: | Front: | | | | |

Snow

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Front: | Front: | Front: | | | | |

Temporal Misalignment

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Multi-View: | Multi-View: | Multi-View: | | | | |

Spatial Misalignment

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Multi-View: | Multi-View: | Multi-View: | | | | |

Beams Reducing

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Front: | Front: | Front: | | | | |

Brightness

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | Multi-View: | Multi-View: | Multi-View: | | | | |

Dark

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | Multi-View: | Multi-View: | Multi-View: | | | | |

Fog

| Severity Level 1 | Severity Level 2 | Severity Level 3 | |-------------------|-------------------|-------------------| | BEV: | BEV: | BEV: | | | | | | Multi-View: | Multi-View: | Multi-View: | | | | |

Note: Right click and click on Open Image in new tab to enlarge an animation

News

• [11.04.2025] v0.0.10 Added model [MoME](https://github.