Open-World / Open-Vocabulary From survey PDF — verify links
Anomize
Anomize — Novel-Category Alignment
CVPR · 2025Human-reviewed summary status
Survey summary
Anomize targets two challenges in open-vocabulary VAD -- detection ambiguity and categorization confusion -- using multi-level visual data with matching textual information to score novel anomalies, plus label-relation-guided text encoding that better aligns novel videos with their labels.
Taxonomy placement
Open-World
Open-WorldOpen-VocabularyNovel-Category Alignment
Method characteristics
Open-World / Open-VocabularyDual-streamGroup-guided text encoding