A benchmark dataset for evaluating spatio-temporal reasoning during vision-and-language navigation tasks, used with the LASAR model. The dataset was created by TMarcus and is associated with a CVPR 2026 publication. It includes video files and associated data files organized in a directory structure.
Use Cases
- Benchmarking navigation agents on spatio-temporal reasoning based on the dataset's described purpose.
- Training models for vision-and-language navigation based on the provided video and data files.
- Research on latent cognitive map representations based on the associated LASAR publication.
Strengths
- Associated with a peer-reviewed CVPR 2026 publication, providing academic credibility.
- Designed for a specific, advanced AI task (spatio-temporal reasoning in navigation).
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 dataset scale are unknown, which may limit suitability assessment.
Provenance
- Source
- huggingface
- Freshness
- Last updated 2026-06-02 18:06:20; freshness should be verified.