Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Lei, Jingyu, Wang, Gaoang, Lee, Der-Horng
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2511.14072
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866917088396836864
author Lei, Jingyu
Wang, Gaoang
Lee, Der-Horng
author_facet Lei, Jingyu
Wang, Gaoang
Lee, Der-Horng
contents Large Vision-Language Models (LVLMs) usually suffer from prohibitive computational and memory costs due to the quadratic growth of visual tokens with image resolution. Existing token compression methods, while varied, often lack a high-level semantic understanding, leading to suboptimal merges, information redundancy, or context loss. To address these limitations, we introduce CORE (Compact Object-centric REpresentations), a new paradigm for visual token compression. CORE leverages an efficient segmentation decoder to generate object masks, which serve as a high-level semantic prior to guide the merging of visual tokens into a compact set of object-centric representations. Furthermore, a novel centroid-guided sorting mechanism restores a coherent spatial order to the merged tokens, preserving vital positional information. Extensive experiments show that CORE not only establishes a new state-of-the-art on six authoritative benchmarks for fixed-rate compression, but also achieves dramatic efficiency gains in adaptive-rate settings. Even under extreme compression, after aggressively retaining with only 2.2% of all visual tokens, CORE still maintains 97.4% of baseline performance. Our work demonstrates the superiority of object-centric representations for efficient and effective LVLM processing.
format Preprint
id arxiv_https___arxiv_org_abs_2511_14072
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CORE: Compact Object-centric REpresentations as a New Paradigm for Token Merging in LVLMs
Lei, Jingyu
Wang, Gaoang
Lee, Der-Horng
Computer Vision and Pattern Recognition
Large Vision-Language Models (LVLMs) usually suffer from prohibitive computational and memory costs due to the quadratic growth of visual tokens with image resolution. Existing token compression methods, while varied, often lack a high-level semantic understanding, leading to suboptimal merges, information redundancy, or context loss. To address these limitations, we introduce CORE (Compact Object-centric REpresentations), a new paradigm for visual token compression. CORE leverages an efficient segmentation decoder to generate object masks, which serve as a high-level semantic prior to guide the merging of visual tokens into a compact set of object-centric representations. Furthermore, a novel centroid-guided sorting mechanism restores a coherent spatial order to the merged tokens, preserving vital positional information. Extensive experiments show that CORE not only establishes a new state-of-the-art on six authoritative benchmarks for fixed-rate compression, but also achieves dramatic efficiency gains in adaptive-rate settings. Even under extreme compression, after aggressively retaining with only 2.2% of all visual tokens, CORE still maintains 97.4% of baseline performance. Our work demonstrates the superiority of object-centric representations for efficient and effective LVLM processing.
title CORE: Compact Object-centric REpresentations as a New Paradigm for Token Merging in LVLMs
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2511.14072