Reproduce “Visual Affordance Prediction: Survey and Reproducibility” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Tommaso Apicella, Alessio Xompero, Andrea Cavallaro
Researchers from IIT, Queen Mary University of London, and Idiap/EPFL developed a unified formulation for visual affordance prediction and introduced the "Affordance Sheet" to promote reporting standards. This work systematically reviews existing methodologies and datasets, critically identifying pervasive reproducibility challenges across the field.
Key insights
- Existing visual affordance prediction methods commonly provide only partial solutions by addressing only a subset of the proposed "what, where, how" aspects.
- Current datasets exhibit significant limitations, including small scale, narrow diversity, reliance on controlled environments, and inconsistent or ambiguous annotations, hindering generalization.
- A pervasive reproducibility crisis affects the field, stemming from the scarcity of public code, trained models, detailed experimental setups, and consistent performance metrics.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.