Reproduce “MetaCOG: A Hierarchical Probabilistic Model for Learning Meta-Cognitive Visual Representations” 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
Marlene D. Berke, Zhangir Azerbayev, Mario Belledonne, Zenna Tavares, Julian Jara-Ettinger
MetaCOG combines detections from multiple views and known camera poses to infer persistent scene objects and category-specific detector misses and hallucinations. It learns detector error tendencies by comparing views under an object-persistence assumption. On synthetic test scenes, its 2D Jaccard accuracy was 0.76, versus 0.69 for Faster R-CNN output and 0.58 for a lesion; results do not establish real-world reliability.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.