JESTER: Event-Based Hand Gesture Recognition Dataset

JESTER is a large-scale, densely labeled video dataset depicting humans performing predefined hand gestures in front of webcams. It was originally created by crowd workers and is widely used for training robust machine learning models for gesture recognition. This specific version is a custom event-based conversion of a subset of the original dataset, uploaded by Ishara5 on Hugging Face in May 2025.

Use Cases

• Train gesture recognition models based on densely labeled video data.
• Benchmark event-based vision algorithms based on the converted event-based format.
• Develop robust models for human-computer interaction based on predefined hand gestures.

Strengths

• Based on the original JESTER dataset, which is described as large-scale and densely labeled.
• Specifically converted for event-based vision, a niche and growing research area.

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 and file size are unknown, which may limit suitability assessment.

Provenance

Source

Original JESTER dataset created by crowd workers; converted subset by Ishara5.

Collection Method

Likely involves conversion of standard video frames to an event-based representation.

Freshness

Last updated 2025-05-08 05:47:42; freshness should be verified.