HAR-Bi: Enhanced Human Action Recognition Dataset from Combined Sources
Available on 1 platform
Sign in to view source links and access this dataset
Description
An enhanced dataset for Human Action Recognition, created by combining two existing Kaggle datasets. The dataset is hosted on Kaggle, but the specific author, organization, and creation date are not provided. Its size, row count, and file formats are unspecified.
Use Cases
Train action classification models based on the enhanced video or image data.
Benchmark model performance against a combined dataset to test generalization.
Develop pose estimation or temporal activity detection pipelines using the action recognition labels.
Strengths
Dataset is described as an 'enhanced' version, suggesting potential improvements over the original sources.
Combines data from two distinct Kaggle datasets, which may increase variety or scale.
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 is unknown, which may limit suitability assessment.
Provenance
Source
Kaggle (combined from two unspecified Kaggle datasets)
Collection Method
Combination and enhancement of existing datasets.
License is unknown; users must verify terms before use.