ADOC

A Day on Campus Dataset. An Anomaly Detection Dataset for Events in a Single Camera.

Anomaly Event Examples

Visual examples from the ADOC dataset showing diverse events under varying conditions.

Golf Cart Anomaly

Golf Cart on Walkway

Person on Grass Anomaly

Person on Grass

Skateboard Anomaly

Person on Skateboard

Conversation Event

Having a Conversation

Bike Anomaly

Riding Bike (Day)

Umbrella Event

Person with Umbrella

Mobility Scooter Anomaly

Mobility Scooter

Night Bike Anomaly

Riding Bike (Night)

The ADOC dataset is a foundational resource for anomaly detection in single-camera surveillance. We acquired 24 continuous hours of 1080p, H.264 video from a live university campus. The data encapsulates extreme variations in illumination, background clutter, and crowd density.

We define anomalies as statistically “rare” events (low-frequency occurrences). Our data is meticulously annotated with 875 events, including high-frequency actions and specific rare activities like “golf cart on walkway” or “person with a mobility scooter.”

Selected Anomaly Events

Event Name Frequency
Person riding a bike 238
Person on grass 132
Golf cart on walkway 105
Having a conversation 58
Person on skateboard 48

Project Resources

References

2020

  1. A day on campus-an anomaly detection dataset for events in a single camera
    Mantini Pranav, Li Zhenggang, and others
    In Proceedings of the Asian Conference on Computer Vision, 2020