CRAFT combines ConceptNet or LLM-derived affordance priors with CLIP visual similarity in an iterative energy-based re-ranking loop, achieving the best non-oracle accuracy on a functional affordance grounding benchmark.
The theory of affordances:(1979)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
CRAFT: A Neuro-Symbolic Framework for Visual Functional Affordance Grounding
CRAFT combines ConceptNet or LLM-derived affordance priors with CLIP visual similarity in an iterative energy-based re-ranking loop, achieving the best non-oracle accuracy on a functional affordance grounding benchmark.