Research map / Vision-language and multimodal / Multimodal LLMs
CoverPrune: Coverage-Driven Token Pruning for 3D VLMs via Optimal Transport
Peng Ling, Yingda Yin, Lingting Zhu, Weikai Chen, Shengju Qian, Zeyu HU, Xin Wang, Wenming Yang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add feature cost term to FST transport cost (ablation) | CoverPrune (GS-Reasoner) · VSI-Bench R=20% | Overall accuracy: 56.18 -> 59.76 | Improved |
| add geometry cost term (ablation) | CoverPrune (GS-Reasoner) · VSI-Bench R=20% | Overall accuracy: 59.02 -> 59.76 | Improved |
| add temporal cost term (ablation) | CoverPrune (GS-Reasoner) · VSI-Bench R=20% | Overall accuracy: 58.96 -> 59.76 | Improved |
| add FST capacity weighting (ablation) | CoverPrune (GS-Reasoner) · VSI-Bench R=20% | Overall accuracy: 59.59 -> 59.76 | Improved |
| use CoverPrune instead of DTC at 10% retention | GS-Reasoner · VSI-Bench R=10% | Avg accuracy: 51.66 -> 56.83 | Improved |
| use CoverPrune instead of VisionZip at 5% retention | GS-Reasoner · VSI-Bench R=5% | Avg accuracy: 46.10 -> 52.66 | Improved |
| use CoverPrune instead of DTC at 10% retention | VLM-3R · VSI-Bench R=10% | Avg accuracy: 52.64 -> 54.69 | Improved |
| use CoverPrune instead of DTC at 5% retention | VLM-3R · VSI-Bench R=5% | Avg accuracy: 49.83 -> 51.28 | Improved |
| use CoverPrune at 20% retention vs full tokens | GS-Reasoner · VSI-Bench | Avg accuracy: 64.70 -> 59.76 | Hurt |
| use CoverPrune instead of FastVID at 10% retention | GS-Reasoner · Scan2Cap R=10% | CIDEr: 85.68 -> 93.84 | Improved |
| use CoverPrune-Lite instead of CoverPrune | GS-Reasoner · VSI-Bench R=10% | Avg accuracy: 56.83 -> 56.94 | No clear effect |
| use CoverPrune-Lite instead of CoverPrune for pruning time | GS-Reasoner · VSI-Bench efficiency | Pruning time (s): 2.53 -> 0.41 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VisionZip — VisionZip: Longer is Better but Not Necessary in Vision Language Models (CVPR 2025) · better on 4 of 4 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- VLM-Pruner: Buffering for Spatial Sparsity in an Efficient VLM Centrifugal Token Pruning Paradigm (CVPR 2026 · similarity 0.90)
- SwiftVLM: Efficient Vision-Language Model Inference via Cross-Layer Token Bypass (NeurIPS 2026 · similarity 0.89)
- IVC-Prune: Revealing the Implicit Visual Coordinates in LVLMs for Vision Token Pruning (ICLR 2026 · similarity 0.89)
- QuietPrune: Query-Guided Early Token Pruning for Vision-Language Models (CVPR 2026 · similarity 0.89)
- Beyond 3D VQAs: Injecting 3D Spatial Priors into Vision-Language Models for Enhanced Geometric Reasoning (CVPR 2026 · similarity 0.88)
- TopV: Compatible Token Pruning with Inference Time Optimization for Fast and Low-Memory Multimodal Vision Language Model (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
