The key design insight is hierarchical ordering: by first compressing spatial dimensions, VETO drastically reduces the cost of subsequent global temporal matching, bypassing the efficiency wall of single-axis approaches, with an advantage that grows with modern fully-fused attention infrastructure.
Abstract
Processing long videos with Vision-Language Models (VLMs) is bottlenecked by the quadratic cost of visual tokens, making long-form inference prohibitively expensive. While single-axis compression methods mitigate this, they hit a hard efficiency floor because they treat spatial and temporal redundancy independently. We present VETO (Video Efficient Token Optimization for Vision-Language Models), a training-optional plug-in that eliminates this bottleneck through dual-axis compression: (i) an intra-frame compressor that merges semantically similar tokens within each frame via optimal-transport inspired matching, and (ii) an inter-frame compressor that identifies and merges temporally redundant frames. The key design insight is hierarchical ordering: by first compressing spatial dimensions, VETO drastically reduces the cost of subsequent global temporal matching, bypassing the efficiency wall of single-axis approaches, with an advantage that grows with modern fully-fused attention infrastructure. Empirically, VETO achieves up to 45% faster inference (e.g., on LLaVA-OneVision-7B) while preserving or improving accuracy. Under extreme token starvation (10% budget), VETO outperforms VFlowOpt (54.9%), VisionZip (52.6%), and FastV (47.9%) with 55.7% accuracy. We demonstrate universal applicability across LLaVA-OneVision, InternVL-2.5, and LongVA, with zero-shot accuracy preserved or improved in all cases.
Token-Budget Distillation (TBD) is presented, a parameter-efficient fine-tuning framework for adapting video VLMs under a fixed token budget and consistently outperforms compression-only baselines under both moderate and aggressive compression.
Xiao-Yang Guo, Guoping Luo, Jusheng Zhang et al.· 1 citation
Experiments on FastVLM-1.5B across multiple vision-language benchmarks demonstrate that VPRune achieves a favorable accuracy--compression trade-off, with particularly pronounced advantages under aggressive compression, demonstrating its practicality for resource-constrained LVLM deployment.
Vision-Language-Action (VLA) models present a promising paradigm for general-purpose robot manipulation, but closed-loop deployment is often limited by inference efficiency: at each timestep, high-resolution, often multi-view images are encoded into hundreds of visual tokens that dominate transformer computation and la...
Jiu-Yi Xu, Cheng-Yuan Qian, Zheng-Zhong Tu et al.· 0 citations
Vision-language models (VLMs) have demonstrated impressive capabilities but suffer from substantial computational overhead, as vision tokens dominate the input sequence. This motivates vision token compression as a key direction to alleviate the burden. However, with the emergence of hybrid architectures incorporating...
Hao Jiang, Yi-Ru Mao, Tian-Peng Bu et al.· Proceedings of the 2026 Conf...· 0 citations
PACE (Pixel-Adaptive Condense and Extract), a training-free inference framework that accelerates both the vision encoder and the Large Language Model (LLM) via a unified Condense-and-Extract paradigm, is proposed.
Braco is a lightweight four-step coder that combines transform-basis truncation, input-independent basis-coordinate embeddings, budget-dependent orthogonal re-parameterization, and learned spatial residual tokens from lightweight pooling that forms the favorable empirical accuracy-efficiency frontier under compression.
Rui-Lian Zhong, Yu Li, Zhe-Yu Yan et al.· arXiv.org· 0 citations