bhatvineetMIT

BOPASK-Test

Human-verified evaluation benchmark for spatial-reasoning visual question-answering on robotic grasping and pose estimation tasks. Contains 934 question-answer pairs across core (HANDAL, HOPE, YCB-V) and lab (home/in-the-wild) testsets with RGB images, depth maps, and segmentation masks.

Downloads11

Technical Profile

Modalities
rgbdepthsegmentation
Environment
labhome
Task Types
graspingpose_estimationvisual_question_answeringspatial_reasoningtrajectory_predictiondepth_estimationobject_rearrangement
Data Format
JSON
License
MIT
Part of the BOPASK-Test family

Access

Need custom rgb data?

Claru builds purpose-built datasets for lab applications with dense human annotations and quality assurance.

Request a Sample Pack

Related Datasets