MicroG-HAR-train-ready: Preprocessed Video Dataset for Human Activity Recognition

MicroG-HAR-train-ready is a dataset converted from the MicroG-4M dataset for human activity recognition. The dataset is formatted to be used directly as input data directories for the PySlowFast_for_HAR framework without additional preprocessing. It was authored by lei-qi-233 and last updated on June 9, 2025.

Use Cases

• Training action detection models based on the preprocessed video format.
• Benchmarking HAR algorithms based on the AVA dataset organizational structure.
• Evaluating model performance on a ready-to-use video dataset without preprocessing overhead.

Strengths

• Formatted for direct use with the PySlowFast_for_HAR framework, reducing setup time.
• Follows the organizational format of the established AVA dataset, suggesting a standardized structure.

Limitations

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

Provenance

Source

Converted from the MicroG-4M dataset.

Collection Method

Conversion process described in the author's paper and GitHub repository.

Time Range

null

Freshness

Last updated 2025-06 09 15:14:45; freshness should be verified.

Geography

null