← all datasets

SpatialMQA

Emerging
3papers using it
2024first seen

SpatialMQA is a human-annotated spatial relation reasoning benchmark based on COCO2017, containing 5,392 samples, used to evaluate the ability of multimodal large language models to understand spatial relations in images.

Papers using SpatialMQA (3)

SpatialMQA β€” datasets β€” multimodal