GeoTrace: Geometry-Aware Trajectory Token Compression for Video Large Language Models
Quick Answer
GeoTrace introduces a training-free spatiotemporal token compression framework for Video LLMs, achieving a 12.99× TFLOPs reduction with only 10% visual tokens retained while preserving 99.1% of performance on LLaVA-OneVision.
Quick Take
GeoTrace introduces a training-free spatiotemporal token compression framework for Video LLMs, achieving a 12.99× TFLOPs reduction with only 10% visual tokens retained while preserving 99.1% of performance on LLaVA-OneVision. This method enhances efficiency and robustness in video understanding across various benchmarks and model architectures.
Key Points
- GeoTrace decomposes video evidence into skeleton and residual event tokens.
- Contextual Farthest-Point Anchoring preserves salient and context-consistent tokens.
- Trajectory-Constrained Residual Condensation reduces ambiguity in event tokens.
- Evaluated on four Video LLMs, demonstrating effectiveness across benchmarks.
- Code available for implementation and further research.
Paper Resources
📖 Reader Mode
~2 min readAbstract:Although Video Large Language Models (Video LLMs) have shown strong performance in video understanding, their efficiency is still limited by the large number of visual tokens. Existing video token compression methods typically rely on frame-wise saliency or heuristic token merging, which can over-focus on locally salient regions and produce ambiguous fused features. To address these issues, we propose GeoTrace, a training-free spatiotemporal token compression framework that decomposes video evidence into exact skeleton tokens and traceable residual event tokens. Specifically, Contextual Farthest-Point Anchoring (CFPA) preserves salient, context-consistent, and high-coverage skeleton tokens, while Trajectory-Constrained Residual Condensation (TCRC) compresses residual tokens through one-to-one temporal trajectories and constrained near-manifold condensation, producing traceable event tokens with reduced ambiguity. We evaluate GeoTrace on four Video LLMs across four video understanding benchmarks, and the results demonstrate its effectiveness and generalization across different model architectures and scenarios. On LLaVA-OneVision, with only 10\% visual tokens retained, GeoTrace achieves a \(12.99\times\) TFLOPs reduction while preserving 99.1\% of the vanilla performance. Overall, GeoTrace offers a compact and traceable token representation for efficient and robust Video LLM inference. Code is available at \href{this https URL}{\texttt{Code}}.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2607.09080 [cs.CV] |
| (or arXiv:2607.09080v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2607.09080 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Guohuan Xie [view email]
[v1]
Fri, 10 Jul 2026 03:58:26 UTC (8,177 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.