CVPR Findings2026
Efficient3D : A Unified Framework for Adaptive and Debiased Token Reduction in 3D MLLMs
Yuhui Lin, Siyue Yu, Yuxing Yang, Guangliang Cheng, Jimin Xiao
Abstract
Recent advances in Multimodal Large Language Models (MLLMs) have expanded reasoning capabilities into 3D domains, enabling fine-grained spatial understanding. However, the substantial size of 3D MLLMs and the high dimensionality of input features introduce considerable inference overhead, which limits practical deployment on resource constrained platforms. To overcome this limitation, this paper presents Efficient3D, a unified framework for visual token pruning that accelerates 3D MLLMs while maintaining competitive accuracy. The proposed framework introduces a Debiased Visual Token Importance Estimator (DVTIE) module, which considers the influence of shallow initial layers during attention aggregation, thereby producing more reliable importance predictions for visual tokens. In addition, an Adaptive Token Rebalancing (ATR) strategy is developed to dynamically adjust pruning strength based on scene complexity, preserving semantic completeness and maintaining balanced attention across layers. Together, they enable context-aware token reduction that maintains essential semantics with lower computation. Comprehensive experiments conducted on five representative 3D vision and language benchmarks, including ScanRefer, Multi3DRefer, Scan2Cap, ScanQA, and SQA3D, demonstrate that Efficient3D achieves superior performance compared with unpruned baselines, with a +2.57% CIDEr improvement on the Scan2Cap dataset. Therefore, Efficient3D provides a scalable and effective solution for efficient inference in 3D MLLMs. The code is released at: https://github.com/sol924/Efficient3D
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
- Chat-Scene: Bridging 3D Scene and Large Language Models with Object Identifiers cites Multi3DRefer: Grounding Text Description to Multiple 3D Objects
- Chat-Scene: Bridging 3D Scene and Large Language Models with Object Identifiers cites An Embodied Generalist Agent in 3D World
Builds on45
- DynamicViT: Efficient Vision Transformers with Dynamic Token SparsificationYongming Rao, Wenliang Zhao, Benlin Liu, Jiwen Lu et al.NeurIPS 2021 · 1,343 citations
- 3D-LLM: Injecting the 3D World into Large Language ModelsYining Hong, Haoyu Zhen, Peihao Chen, Shuhong Zheng et al.NeurIPS 2023 · 662 citations
- An Embodied Generalist Agent in 3D WorldJiangyong Huang, Silong Yong, Xiaojian Ma, Xiongkun Linghu et al.ICML 2024 · 361 citations
- Chat-Scene: Bridging 3D Scene and Large Language Models with Object IdentifiersHaifeng Huang, Yilun Chen, Zehan Wang, Rongjie Huang et al.NeurIPS 2024 · 230 citations
- Multi3DRefer: Grounding Text Description to Multiple 3D ObjectsYiming Zhang, ZeMing Gong, Angel X. ChangICCV 2023 · 157 citations
Related papers
- Fast3D: Accelerating 3D Multi-modal Large Language Models for Efficient 3D Scene UnderstandingWencan Huang, Daizong Liu, Wei HuACM MM 2025 · 1 citation
- ST3: Accelerating Multimodal Large Language Model by Spatial-Temporal Visual Token TrimmingJiedong Zhuang, Lu Lu, Ming Dai, Rui Hu et al.AAAI 2025 · 1 citation
- Geometry-Guided 3D Visual Token Pruning for Video-Language ModelsHan Li, Zehao Huang, Jiahui Fu, Naiyan Wang et al.CVPR 2026 · 3 citations
- Don't Just Chase "Highlighted Tokens" in MLLMs: Revisiting Visual Holistic Context RetentionXin Zou, Di Lu, Yizhou Wang, Yibo Yan et al.NeurIPS 2025 · 49 citations
- D²Pruner: Debiased Importance and Structural Diversity for MLLM Token PruningEvelyn Zhang, Fufu Yu, Aoqi Wu, Zichen Wen et al.AAAI 2026 · 1 citation
