CarlAnomaly

A CARLA-based Dataset for Multi-Modal Anomaly Detection
in Autonomous Driving Scenarios

Multi-Modal

  • 4 RGB-D Cameras
  • LiDAR
  • IMU
  • GNSS

Anomalies

  • Various Types
  • Manually Verified
  • Diverse Environmental Conditions

Large Scale

  • 2 hours of training data
  • 1 hour of validation data
  • 500 x 20s clips for testing
  • 1.3 TB
Paper
@inproceedings{


}
Example Anomalies

CarlAnomaly features a wide range of anomalies.

Overall, there are x different types of anomalies:

  • Randomly appearing and disappearing unknown objects
  • Traffic Lights Malfunctioning in several ways (off, blinking single color, switching to fast)
  • Instantaneous weather changes
  • Spawning or vanishing objects
License: MIT
The CarlAnomaly dataset, including all software and models, is licensed under the MIT which you can find here.