Overhead Traffic Anomalies: 32 Anomaly Categories from Static Cameras

32 distinct anomaly categories provide a structured framework for severity-aware analysis of traffic incidents. Developed by H. Lichtenberg at Starwit Technologies GmbH in 2025, this dataset contains traffic anomalies captured from static overhead cameras at intersections and roundabouts. The data supports the author's master thesis on probabilistic forecasting for anomaly detection in traffic applications.

Use Cases

• Training anomaly detection models based on 32 labeled anomaly categories.
• Developing severity-aware traffic monitoring systems based on the relevance mapping for anomaly severity.
• Benchmarking probabilistic forecasting approaches for traffic incidents using object tracking data.
• Analyzing intersection and roundabout safety based on categorized anomaly events.

Strengths

• Contains 32 specifically defined anomaly categories for detailed classification.
• Includes a relevance mapping designed for severity-aware analysis of incidents.
• Originates from a focused master thesis project on probabilistic forecasting in traffic.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic and temporal bias inherent to the specific camera locations used.

Provenance

Source

Starwit Technologies GmbH

Collection Method

Captured from static overhead cameras at intersections and roundabouts, likely involving object tracking.

Freshness

Last updated 2025-09-12 12:53:05.