TAG-Bench: 828 Generated Human Action Videos with Human Ratings

828 generated video clips of human actions form the TAG-Bench benchmark for evaluating human motion realism in video generative models. The dataset includes human ratings collected for these clips. It was created by author dghadiya and was last updated on the Hugging Face platform in May 2026.

Use Cases

• Benchmarking video generation models based on human ratings of motion realism.
• Training generative models to produce more realistic human actions based on the rating feedback.
• Analyzing the correlation between specific video features and perceived human motion quality.

Strengths

• Contains 828 video clips, providing a substantial sample size for evaluation.
• Includes human ratings, offering a direct measure of perceived realism for the generated content.
• Released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license, facilitating open use.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Freshness should be verified as the last update timestamp is in the future (2026-05-06).

Provenance

Source

Hugging Face dataset uploaded by author dghadiya.

Collection Method

Likely contains videos generated by AI models and ratings collected from human evaluators.

Freshness

Last updated 2026-05-06 19:15:15