Huperflow: Human Perceived Optical Flow Derived from Benchmark Images

Human perceived optical flow and motion data derived from images obtained from several third-party benchmark datasets. The data is intended for research and non-commercial use only. The dataset was last updated on March 25, 2025, by author sunana.

Use Cases

• Training optical flow estimation models based on human-perceived motion labels.
• Benchmarking computer vision algorithms against human perception of motion.
• Studying the gap between computed and human-perceived optical flow.
• Developing models for tasks requiring an understanding of motion in images.

Strengths

• Derived from established third-party benchmark datasets, suggesting a foundation of validated imagery.
• Focuses on human-perceived motion, a less common annotation type than synthetic flow.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Multiple third-party benchmark datasets (specific sources not named).

Collection Method

Derived from images and labels from publicly available datasets.

Freshness

Last updated 2025-03-25 00:58:35; freshness should be verified.