Reproduce “Open-World Hierarchical Perception: Taxonomic Abstraction over Class-Agnostic Proposals for the Safe Handling of Out-of-Vocabulary Road Objects” 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
Felix Schaller
The authors introduce HOWC, a hierarchical road-object classifier that pools CLIP scores from specific labels into broader taxonomy categories and returns UNKNOWN OBSTACLE when evidence is insufficient. On 235 ground-truth crops from a COCO leave-classes-out test, HOWC made 0% confident wrong specific-label predictions and safely handled 94% under the study’s definition, mostly by abstaining; it still chose wrong super-categories 6%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.