Violence vs Non-Violence Human Interactions from Video Frames

Annotated interaction data extracted from video frames for detecting violent vs non-violent human interactions. The dataset, created by harmesh95, was last updated on August 17, 2025. It includes bounding boxes, pose keypoints, motion features, and violence indicators for pairs of interacting persons.

Use Cases

• Train a binary classifier for violence detection based on annotated interaction labels.
• Develop models for human pose analysis based on the pose keypoints mentioned in the description.
• Analyze motion patterns in human interactions based on the described motion features.
• Benchmark object detection models using the bounding box data for interacting persons.

Strengths

• Data is explicitly annotated for violent and non-violent interactions, providing clear labels.
• Includes multiple feature types such as bounding boxes, pose keypoints, and motion features for a multi-modal analysis approach.

Limitations

• Row count, file size, and column-level documentation are unknown, which limits suitability assessment.
• Description metadata is limited; actual data quality and feature specifics require manual inspection after download.

Provenance

Source

huggingface user harmesh95

Collection Method

Extracted and annotated from video frames.

Time Range

null

Freshness

Last updated 2025-08-17 19:09:35.

Geography

null