TUM Traffic Accident Dataset

Paper Model Dataset Labeling Tool Development Kit

Overview

The TUM Traffic Accident (TUMTraf-A) dataset is the first high-quality real-world accident dataset for the 3D object detection, segmentation and tracking task in autonomous driving.

It contains:
  • data collected by 5 sensors (cameras and LiDARs) simultaneously from onboard and roadside sensors.
  • 8,944 labeled frames.
  • Labeled 3D bounding boxes, instance segmentation masks, trajectories, and track IDs.
  • 10 real accidents including overturning trailers, vehicles catching fire and collisions.
  • HD map of the highway.
  • Labels in OpenLABEL standard.
  • A dataset development kit to load, preprocess, visualize, convert labels, and to evaluate accident detection models.

Abstract

Even though a significant amount of work has been done to increase the safety of transportation networks, accidents still occur regularly. They must be understood as an unavoidable and sporadic outcome of traffic networks. We present the TUM Traffic Accident (TUMTraf-A) dataset, a collection of real-world highway accidents. It contains ten sequences of vehicle crashes at high-speed driving with labeled 3D boxes and track IDs within 8,944 labeled frames recorded from four roadside cameras and LiDARs at 10 Hz. The dataset contains ten object classes and is provided in the OpenLABEL format. We propose an accident detection model that combines a rule-based approach with a learning-based one. Experiments and ablation studies on our dataset show the robustness of our proposed method. The dataset, model, and code are available on our project website.

Sensor Setup

The following roadside sensors were used:
  • 4x Basler ace acA1920-50gc, 1920×1200, Sony IMX174 with 16 mm and 50 mm lenses, frame rate: 50 Hz
  • 1x Valeo LiDAR (SCALA B2), 16 vertical layers, 133° horiz. FOV, 0.125° x 0.6° angular resolution, 200 m range (@80% reflectivity), frame rate: 25 Hz
infrastructure_sensors

Visualization of roadside sensors used to record the TUM Traffic Accident Dataset infrastructure perspective.

Benchmark

Config BEV mAP 3D mAP
Domain Modality   Easy Moderate Hard Average
TBD TBD TBD TBD TBD TBD TBD
Evaluation results (BEV mAP and 3D mAP) of AccidentDet3D on our
TUM Traffic Accident test set.

Acknowledgements

This research was supported by the Federal Ministry of Education and Research in Germany within the AUTOtech.agil project, Grant Number: 01IS22088U.