Optical Flow Pedestrian Dataset from UCSD

The UCSD Optical Flow Pedestrian Dataset contains optical flow fields computed from surveillance videos to study pedestrian motion patterns in crowded scenes. It supports research on crowd behavior analysis, anomaly detection.

Use Cases

• Train optical flow networks on pedestrian movements (verify after download)
• Benchmark crowd motion prediction algorithms (verify after download)

Strengths

• Optical flow fields precomputed for consistency
• Focus on crowded pedestrian scenarios

Limitations

• Metadata completeness unknown
• Last update date unknown

Provenance

Source

UCSD

Collection Method

Optical flow computed from surveillance footage

Time Range

Unknown

Geography

UCSD campus