{"slug": "cover-coverage-based-token-pruning-for-multi-view-3d-reasoning-in-vlms", "title": "CoVeR: Coverage-Based Token Pruning for Multi-View 3D Reasoning in VLMs", "summary": "A new method called CoVeR (Coverage-Based Token Pruning) reduces the number of visual tokens processed by 2D vision-language models (VLMs) when reasoning about 3D scenes from multi-view images, addressing the high computational cost of thousands of redundant tokens. The approach, detailed in a research paper, aims to improve efficiency of 3D reasoning without sacrificing accuracy.", "body_md": "Representing a 3D scene as multi-view images allows 2D VLMs to reason in 3D by reusing priors from pre-training, sidestepping the scarcity of annotated 3D data. However, it produces thousands of redundant visual tokens whose cost grows with every view. Existing visual token pruners fall into two fam", "url": "https://wpnews.pro/news/cover-coverage-based-token-pruning-for-multi-view-3d-reasoning-in-vlms", "canonical_source": "https://aiflash.com/news/116136/", "published_at": "2026-09-09 06:30:24+00:00", "updated_at": "2026-09-09 06:58:57.084900+00:00", "lang": "en", "topics": ["artificial-intelligence", "computer-vision", "large-language-models", "ai-research"], "entities": ["CoVeR"], "alternates": {"html": "https://wpnews.pro/news/cover-coverage-based-token-pruning-for-multi-view-3d-reasoning-in-vlms", "markdown": "https://wpnews.pro/news/cover-coverage-based-token-pruning-for-multi-view-3d-reasoning-in-vlms.md", "text": "https://wpnews.pro/news/cover-coverage-based-token-pruning-for-multi-view-3d-reasoning-in-vlms.txt", "jsonld": "https://wpnews.pro/news/cover-coverage-based-token-pruning-for-multi-view-3d-reasoning-in-vlms.jsonld"}}