A new multi-modal dataset called the In-Car Sign Language (ICSL) corpus has been introduced to address the challenges of using sign language within shared mobility services like taxis and ride-sharing platforms. The resource focuses on Brazilian Sign Language (Libras) to improve public transport accessibility for the Deaf and Hard-of-Hearing community in confined vehicle interiors.
- The dataset includes high-precision laboratory motion capture data to establish a linguistic baseline and real-world multi-modal recordings from 2D cameras and 3D Time-of-Flight sensors.
- It comprises over 1.5 million frames of synchronized streams featuring Libras users across various in-car scenarios.
- The corpus provides gloss annotations for lexical signs and non-lexical elements to support the training and evaluation of deep neural networks.
This resource enables comparative analyses between synthesized avatar animations and real signing videos, providing a foundation for robust sign language recognition models in constrained, occluded environments.