Spatial-DISE is a benchmark dataset designed to evaluate spatial reasoning capabilities in vision-language models. It focuses on aspects of spatial intelligence including 3D perception, spatial transformation, and geometric reasoning across multiple difficulty levels. The dataset was created by TACPS-liv and was last updated on May 9, 2026.
Use Cases
- Benchmarking model performance on 3D perception tasks based on the described focus.
- Evaluating spatial transformation understanding in multimodal AI systems.
- Testing geometric reasoning capabilities across different difficulty levels as mentioned in the description.
Strengths
- Designed to evaluate multiple aspects of spatial intelligence, including 3D perception and geometric reasoning.
- Structured across multiple difficulty levels to provide a graduated assessment.
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 is unknown, which may limit suitability assessment.
Provenance
- Source
- TACPS-liv
- Freshness
- Last updated 2026-05-09 19:25:23; freshness should be verified.