QMoP: Query Guided Mixture-of-Projector for Efficient Visual Token Compression

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Li, Zhongyang, Li, Yaqian, Fang, Faming, Takezoe, Rinyoichi, Bo, Zi-Hao, Qian, Cheng, Guang, Mo, Zhang, Guixu, Long, Kaiwen
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912977701044224
author Li, Zhongyang
Li, Yaqian
Fang, Faming
Takezoe, Rinyoichi
Bo, Zi-Hao
Qian, Cheng
Guang, Mo
Zhang, Guixu
Long, Kaiwen
author_facet Li, Zhongyang
Li, Yaqian
Fang, Faming
Takezoe, Rinyoichi
Bo, Zi-Hao
Qian, Cheng
Guang, Mo
Zhang, Guixu
Long, Kaiwen
contents Multimodal large language models suffer from severe computational and memory bottlenecks, as the number of visual tokens far exceeds that of textual tokens. While recent methods employ projector modules to align and compress visual tokens into text-aligned features, they typically depend on fixed heuristics that limit adaptability across diverse scenarios. In this paper, we first propose Query Guided Mixture-of-Projector (QMoP), a novel and flexible framework that adaptively compresses visual tokens via three collaborative branches: (1) a pooling-based branch for coarse-grained global semantics, (2) a resampler branch for extracting high-level semantic representations, and (3) a pruning-based branch for fine-grained token selection to preserve critical visual detail. To adaptively coordinate these branches, we introduce the Query Guided Router (QGR), which dynamically selects and weights the outputs from different branches based on both visual input and textual queries. A Mixture-of-Experts-style fusion mechanism is designed to aggregate the outputs, harnessing the strengths of each strategy while suppressing noise. To systematically evaluate the effects of Visual Token Compression, we also develop VTCBench, a dedicated benchmark for evaluating the information loss induced by visual token compression. Extensive experiments demonstrate that despite relying on fundamental compression modules, QMoP outperforms strong baselines and delivers significant savings in memory, computation, and inference time.
format Preprint
id arxiv_https___arxiv_org_abs_2603_21232
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle QMoP: Query Guided Mixture-of-Projector for Efficient Visual Token Compression
Li, Zhongyang
Li, Yaqian
Fang, Faming
Takezoe, Rinyoichi
Bo, Zi-Hao
Qian, Cheng
Guang, Mo
Zhang, Guixu
Long, Kaiwen
Computer Vision and Pattern Recognition
Artificial Intelligence
Multimodal large language models suffer from severe computational and memory bottlenecks, as the number of visual tokens far exceeds that of textual tokens. While recent methods employ projector modules to align and compress visual tokens into text-aligned features, they typically depend on fixed heuristics that limit adaptability across diverse scenarios. In this paper, we first propose Query Guided Mixture-of-Projector (QMoP), a novel and flexible framework that adaptively compresses visual tokens via three collaborative branches: (1) a pooling-based branch for coarse-grained global semantics, (2) a resampler branch for extracting high-level semantic representations, and (3) a pruning-based branch for fine-grained token selection to preserve critical visual detail. To adaptively coordinate these branches, we introduce the Query Guided Router (QGR), which dynamically selects and weights the outputs from different branches based on both visual input and textual queries. A Mixture-of-Experts-style fusion mechanism is designed to aggregate the outputs, harnessing the strengths of each strategy while suppressing noise. To systematically evaluate the effects of Visual Token Compression, we also develop VTCBench, a dedicated benchmark for evaluating the information loss induced by visual token compression. Extensive experiments demonstrate that despite relying on fundamental compression modules, QMoP outperforms strong baselines and delivers significant savings in memory, computation, and inference time.
title QMoP: Query Guided Mixture-of-Projector for Efficient Visual Token Compression
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2603.21232