MedPhyGraph2026cc-by-4.0

MedPhyGraph Procedural Support-Graph Data

Name: MedPhyGraph Procedural Support-Graph Data
Creator: MedPhyGraph
Published: 2026-01-01
License: https://creativecommons.org/licenses/by/4.0/
Keywords: scene-graphs, support-relations, counterfactual-evidence, structured-scene-states, scene-graph-maintenance, support-relation-prediction, simulation, digital-twins

Structured procedural training and evaluation data for MedPhyGraph, a method for counterfactual support-graph maintenance in dynamic built-environment digital twins. Contains 533 candidate-edge samples, 54 procedural scene states, and 136 procedural transfer cases.

Downloads29
Episodes533 candidate-edge samples (324 train / 101 validation / 108 test); 54 procedural scene states; 136 procedural transfer cases

Why This Matters for Physical AI

MedPhyGraph provides structured procedural data for training and evaluating methods that understand and maintain support relationships in dynamic environments, which is essential for robots to manipulate and reason about object stability in physical scenes.

Technical Profile

Modalities: scene-graphssupport-relationscounterfactual-evidencestructured-scene-states
Environment: simulationdigital-twins
Task Types: scene-graph-maintenancesupport-relation-prediction
Episodes: 533 candidate-edge samples (324 train / 101 validation / 108 test); 54 procedural scene states; 136 procedural transfer cases
Data Format: JSON
Annotation Types: scene-graphssupport-relationscounterfactual-evidencebefore-after-states
License: cc-by-4.0

Part of the MedPhyGraph family

Access

View on HuggingFace Read Paper

Need custom scene-graphs data?

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

Request a Sample Pack