Metraipolis › archive › dc_f04395da6516

CoVeR: Coverage-Based Token Pruning for Multi-View 3D Reasoning in VLMs — arXiv:2609.08345

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.08345
License (as archived)
CC0-1.0
Archived
2026-10-16T13:59:36.055Z
SHA-256 of the archived text
94b94d461a724f7dff47df0492b2d96bab43e36eff8f558b8891daee91cbacbb
Stable URL of this page
https://metraipolis.com/documento/dc_f04395da6516

How to cite

CoVeR: Coverage-Based Token Pruning for Multi-View 3D Reasoning in VLMs — arXiv:2609.08345 — arXiv (en), https://arxiv.org/abs/2609.08345 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_f04395da6516 (Metraipolis archive dc_f04395da6516, sha256 94b94d461a724f7dff47df0492b2d96bab43e36eff8f558b8891daee91cbacbb)
BibTeX
@misc{dcf04395da6516,
  title = {CoVeR: Coverage-Based Token Pruning for Multi-View 3D Reasoning in VLMs — arXiv:2609.08345},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.08345},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_f04395da6516; sha256 94b94d461a724f7dff47df0492b2d96bab43e36eff8f558b8891daee91cbacbb},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.08345. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

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 families, each limited in the 3D multi-view setting. Learned importance methods rank tokens by attention or encoder features; because redundancy here is fundamentally spatial, they keep near-duplicate tokens from a few prominent regions and leave most of the scene unrepresented. Voxelization methods improve spatial coverage but cannot enforce an exact token budget and saturate as multi-view observations overlap in 3D, capping retention well below the target. We show that spatial coverage is associated with 3D reasoning performance and introduce CoVeR, a deterministic, training-free selector that uses only token coordinates, with no learned signals. CoVeR selects tokens that collectively cover every region of the scene, and solves the limitations of both families: it enforces an exact per-scene budget, breaks the voxelization saturation plateau, and avoids the near-duplicate selections of learned importance. Extensive experiments show CoVeR outperforms prior SOTAs on all three 3D reasoning benchmarks and generalizes as a plug-and-play module tested across four VLMs. Notably, with only $\approx$8% of visual tokens, it preserves 93.5% of full-token performance, surpassing SOTA by 3.9 percentage points on average across benchmarks.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-09 in cs.LG).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2609.08345 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T13:39:43.274Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-09?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).