Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning

• URL: http://arxiv.org/abs/2509.06461v2
• Date: Thu, 11 Sep 2025 15:24:22 GMT
• Title: Focusing by Contrastive Attention: Enhancing VLMs' Visual Reasoning
• Authors: Yuyao Ge, Shenghua Liu, Yiwei Wang, Lingrui Mei, Baolong Bi, Xuanshan Zhou, Jiayu Yao, Jiafeng Guo, Xueqi Cheng,
• Abstract summary: Vision-Language Models (VLMs) have demonstrated remarkable success across diverse visual tasks, yet their performance degrades in complex visual environments.<n>We show that visual complexity strongly correlates with attention entropy, negatively impacting reasoning performance.<n>We propose Contrastive Attention Refinement for Visual Enhancement (CARVE), a training-free method that extracts task-relevant visual signals through attention contrasting at the pixel level.
• Score: 79.34909830834464
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Vision-Language Models (VLMs) have demonstrated remarkable success across diverse visual tasks, yet their performance degrades in complex visual environments. While existing enhancement approaches require additional training, rely on external segmentation tools, or operate at coarse-grained levels, they overlook the innate ability within VLMs. To bridge this gap, we investigate VLMs' attention patterns and discover that: (1) visual complexity strongly correlates with attention entropy, negatively impacting reasoning performance; (2) attention progressively refines from global scanning in shallow layers to focused convergence in deeper layers, with convergence degree determined by visual complexity. (3) Theoretically, we prove that the contrast of attention maps between general queries and task-specific queries enables the decomposition of visual signal into semantic signals and visual noise components. Building on these insights, we propose Contrastive Attention Refinement for Visual Enhancement (CARVE), a training-free method that extracts task-relevant visual signals through attention contrasting at the pixel level. Extensive experiments demonstrate that CARVE consistently enhances performance, achieving up to 75% improvement on open-source models. Our work provides critical insights into the interplay between visual complexity and attention mechanisms, offering an efficient pathway for improving visual reasoning with contrasting attention.

Related papers

• Revealing and Enhancing Core Visual Regions: Harnessing Internal Attention Dynamics for Hallucination Mitigation in LVLMs [67.69730908817321]
  Internal Positive Attention Dynamics (PAD) in LVLMs naturally reveal semantically core visual regions under the distortions of attention sinks.<n>We propose Positive Attention Dynamics Enhancement (PADE), a training-free attention intervention that constructs a PAD map to identify semantically core visual regions.
  arXiv  Detail & Related papers  (2026-02-17T13:08:06Z)
• Capturing Gaze Shifts for Guidance: Cross-Modal Fusion Enhancement for VLM Hallucination Mitigation [8.805397340243557]
  Vision language models (VLMs) often generate hallucination, i.e., content that cannot be substantiated by visual inputs.<n>We propose a method called Gaze Shift-Guided Cross-modal Fusion Enhancement (GIFT) to mitigate hallucination.
  arXiv  Detail & Related papers  (2025-10-24T23:04:26Z)
• Agentic Jigsaw Interaction Learning for Enhancing Visual Perception and Reasoning in Vision-Language Models [63.69856480318313]
  AGILE formulates jigsaw solving as an interactive process, enabling the model to progressively engage with the environment.<n>We show that AGILE substantially boosts performance on jigsaw tasks of varying complexity.<n>We also demonstrate strong generalization across 9 general vision tasks, achieving an average improvement of 3.1%.
  arXiv  Detail & Related papers  (2025-10-01T17:58:05Z)
• Cross-Layer Vision Smoothing: Enhancing Visual Understanding via Sustained Focus on Key Objects in Large Vision-Language Models [13.17978215666921]
  Large Vision-Language Models (LVLMs) can accurately locate key objects in images, yet their attention to these objects tends to be very brief.<n>Motivated by the hypothesis that sustained focus on key objects can improve LVLMs' visual capabilities, we propose Cross-Layer Vision Smoothing (CLVS)<n> CLVS achieves state-of-the-art performance on a variety of visual understanding tasks.
  arXiv  Detail & Related papers  (2025-09-16T09:54:01Z)
• Visual Structures Helps Visual Reasoning: Addressing the Binding Problem in VLMs [9.406760867809124]
  This paper introduces VISER (Visual Input Structure for Enhanced Reasoning), a simple yet effective intervention.<n>We empirically demonstrate substantial performance improvements across core visual reasoning tasks.<n>We find that low-level visual structuring is a powerful and underexplored direction for improving compositional visual reasoning.
  arXiv  Detail & Related papers  (2025-06-27T11:44:40Z)
• Caption This, Reason That: VLMs Caught in the Middle [3.4820139118440676]
  Vision-Language Models (VLMs) have shown remarkable progress in visual understanding in recent years.<n>They still lag behind human capabilities in specific visual tasks such as counting or relational reasoning.<n>We analyze VLM performance along core cognitive axes: Perception, Attention, and Memory.
  arXiv  Detail & Related papers  (2025-05-24T14:25:48Z)
• See What You Are Told: Visual Attention Sink in Large Multimodal Models [4.024850952459758]
  Large multimodal models (LMMs) "see" images by leveraging the attention mechanism between text and visual tokens in the transformer decoder.<n>Recent findings indicate that LMMs have an extraordinary tendency to consistently allocate high attention weights to specific visual tokens.<n>We introduce Visual Attention Redistribution ( VAR), a method that redistributes attention in image-centric heads.
  arXiv  Detail & Related papers  (2025-03-05T09:55:07Z)
• Instruction-Guided Fusion of Multi-Layer Visual Features in Large Vision-Language Models [50.98559225639266]
  We investigate the contributions of visual features from different encoder layers using 18 benchmarks spanning 6 task categories.<n>Our findings reveal that multilayer features provide complementary strengths with varying task dependencies, and uniform fusion leads to suboptimal performance.<n>We propose the instruction-guided vision aggregator, a module that dynamically integrates multi-layer visual features based on textual instructions.
  arXiv  Detail & Related papers  (2024-12-26T05:41:31Z)
• VipAct: Visual-Perception Enhancement via Specialized VLM Agent Collaboration and Tool-use [74.39058448757645]
  We present VipAct, an agent framework that enhances vision-language models (VLMs) VipAct consists of an orchestrator agent, which manages task requirement analysis, planning, and coordination, along with specialized agents that handle specific tasks. We evaluate VipAct on benchmarks featuring a diverse set of visual perception tasks, with experimental results demonstrating significant performance improvements.
  arXiv  Detail & Related papers  (2024-10-21T18:10:26Z)
• Top-Down Visual Attention from Analysis by Synthesis [87.47527557366593]
  We consider top-down attention from a classic Analysis-by-Synthesis (AbS) perspective of vision. We propose Analysis-by-Synthesis Vision Transformer (AbSViT), which is a top-down modulated ViT model that variationally approximates AbS, and controllable achieves top-down attention.
  arXiv  Detail & Related papers  (2023-03-23T05:17:05Z)
• Heterogeneous Contrastive Learning: Encoding Spatial Information for Compact Visual Representations [183.03278932562438]
  This paper presents an effective approach that adds spatial information to the encoding stage to alleviate the learning inconsistency between the contrastive objective and strong data augmentation operations. We show that our approach achieves higher efficiency in visual representations and thus delivers a key message to inspire the future research of self-supervised visual representation learning.
  arXiv  Detail & Related papers  (2020-11-19T16:26:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.