552,992 high-resolution images categorized into 18 distinct hand gesture classes such as 'stop', 'peace', and 'fist'. Each image includes bounding box annotations for hand localization and was captured by 34,730 unique participants in diverse real-world environments.
Use Cases
- Train a gesture classification model using the 18 class labels and image files
- Develop a hand detection and localization system using the bounding box coordinates
- Perform domain adaptation studies by testing models against the diverse background and lighting metadata
Strengths
- 552,992 images across 18 gesture categories including 'like', 'dislike', 'ok', and 'mute'
- Bounding box annotations for every hand instance provided in JSON format
- Data collected from 34,730 unique individuals to ensure high variance in hand morphology
- Images captured in varied lighting conditions and distances ranging from 0.5 to 4 meters