SpaRRTa: 149,145 Synthetic Image-Text Pairs for Spatial Intelligence

SpaRRTa contains 149,145 synthetic paired samples designed to evaluate spatial intelligence in visual foundation models, published by turhancan97 in 2026. The collection features images embedded in Parquet shards alongside detailed metadata describing scene variants and spatial configurations.

Use Cases

• Evaluating spatial reasoning by comparing model outputs against ground truth parameters in raw_params_json
• Fine-tuning vision models to recognize specific spatial configurations defined by the scene and variant columns
• Analyzing model sensitivity to scene changes using the scene_variant paired samples

Strengths

• 149,145 paired samples
• MIT licensed for open research
• Includes raw generation parameters in raw_params_json

Limitations

• Synthetic nature may not capture the full complexity of real-world lighting and physics
• Lack of distinct validation or test splits

Provenance

Source

turhancan97 (Arxiv 2601.11729)

Collection Method

Synthetic generation

Freshness

Last updated March 2026.