20,000 high-definition video recordings of Russian Sign Language (RSL) categorized into 1,000 distinct sign classes. The collection features performances from 194 unique signers captured in diverse environments to support gesture recognition tasks.
Use Cases
- Train video-based classification models for sign language recognition using the class labels
- Evaluate model generalization across different individuals using the signer ID metadata
- Develop real-time gesture detection systems by processing the high-definition video streams
Strengths
- 20,000 video samples provided in 1920x1080 resolution
- 1,000 unique lexical signs from Russian Sign Language
- Data from 194 different signers to account for individual performance variations
- Annotated with class labels and signer identifiers for supervised learning