# Visual Attention Never Fades: Selective Progressive Attention ReCalibration for Detailed Image Captioning in Multimodal Large Language Models

arXiv: https://arxiv.org/abs/2502.01419 · HTML: https://mlsanity.com/paper/2502.01419/

## Key facts

- Cited by 22 papers in the map (data through 14 Oct 2026).
- Placed in Visual token pruning and efficiency, one of 16 directions in Multimodal LLMs / vision-language models.
- Builds on 48 of its 57 references that are in the map.
- First posted on arXiv 3 Feb 2025.

## Placed in

- Multimodal LLMs / vision-language models / [Visual token pruning and efficiency](https://mlsanity.com/mllm/visual-token-pruning-and-efficiency/)

## Builds on (most cited first, 30 of 48)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Llama 2: Open Foundation and Fine-Tuned Chat Models](https://mlsanity.com/paper/2307.09288/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models](https://mlsanity.com/paper/2301.12597/) (2023)
- [GPT-4o System Card](https://mlsanity.com/paper/2410.21276/) (2024)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolution](https://mlsanity.com/paper/2409.12191/) (2024)
- [Qwen Technical Report](https://mlsanity.com/paper/2309.16609/) (2023)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [InternVL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks](https://mlsanity.com/paper/2312.14238/) (2023)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [Evaluating Object Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2305.10355/) (2023)
- [Inference-Time Intervention: Eliciting Truthful Answers from a Language Model](https://mlsanity.com/paper/2306.03341/) (2023)
- [Vision Transformers Need Registers](https://mlsanity.com/paper/2309.16588/) (2023)
- [VILA: On Pre-training for Visual Language Models](https://mlsanity.com/paper/2312.07533/) (2023)
- [Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decoding](https://mlsanity.com/paper/2311.16922/) (2023)
- [Instruction Tuning with GPT-4](https://mlsanity.com/paper/2304.03277/) (2023)
- [Object Hallucination in Image Captioning](https://mlsanity.com/paper/1809.02156/) (2018)
- [OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocation](https://mlsanity.com/paper/2311.17911/) (2023)
- [Hallucination of Multimodal Large Language Models: A Survey](https://mlsanity.com/paper/2404.18930/) (2024)
- [A Survey on Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2402.00253/) (2024)
- [Analyzing and Mitigating Object Hallucination in Large Vision-Language Models](https://mlsanity.com/paper/2310.00754/) (2023)
- [Detecting and Preventing Hallucinations in Large Vision Language Models](https://mlsanity.com/paper/2308.06394/) (2023)
- [NVILA: Efficient Frontier Visual Language Models](https://mlsanity.com/paper/2412.04468/) (2024)
- [Multi-Modal Hallucination Control by Visual Information Grounding](https://mlsanity.com/paper/2403.14003/) (2024)
- [Efficient Large Language Models: A Survey](https://mlsanity.com/paper/2312.03863/) (2023)
- [Paying More Attention to Image: A Training-Free Method for Alleviating Hallucination in LVLMs](https://mlsanity.com/paper/2407.21771/) (2024)
- [Compositional Chain-of-Thought Prompting for Large Multimodal Models](https://mlsanity.com/paper/2311.17076/) (2023)

## Built on by (most cited first, 22 of 22)

- [EvoLMM: Self-Evolving Large Multimodal Models with Continuous Rewards](https://mlsanity.com/paper/2511.16672/) (2025)
- [RSCC: A Large-Scale Remote Sensing Change Caption Dataset for Disaster Events](https://mlsanity.com/paper/2509.01907/) (2025)
- [KVSmooth: Mitigating Hallucination in Multi-modal Large Language Models through Key-Value Smoothing](https://mlsanity.com/paper/2602.04268/) (2026)
- [Uncovering Grounding IDs: How External Cues Shape Multimodal Binding](https://mlsanity.com/paper/2509.24072/) (2025)
- [Thinking Diffusion: Penalize and Guide Visual-Grounded Reasoning in Diffusion Multimodal Language Models](https://mlsanity.com/paper/2604.05497/) (2026)
- [Mitigating Visual Context Degradation in Large Multimodal Models: A Training-Free Decoupled Agentic Framework](https://mlsanity.com/paper/2509.23322/) (2025)
- [StAR: Segment Anything Reasoner](https://mlsanity.com/paper/2603.14382/) (2026)
- [Mitigating Hallucinations in Large Vision-Language Models via Causal Route Gating](https://mlsanity.com/paper/2605.24024/) (2026)
- [Addressing Exacerbated Attention Sink for Source-Free Cross-Domain Few-Shot Learning](https://mlsanity.com/paper/2605.25799/) (2026)
- [Attention-space Contrastive Guidance for Efficient Hallucination Mitigation in LVLMs](https://mlsanity.com/paper/2601.13707/) (2026)
- [Focus, Don't Prune: Identifying Instruction-Relevant Regions for Information-Rich Image Understanding](https://mlsanity.com/paper/2603.22815/) (2026)
- [TraversalBench: Challenging Paths to Follow for Vision Language Models](https://mlsanity.com/paper/2604.10999/) (2026)
- [Steer Where It Matters: Token-Level Visual-Sensitivity Steering for LVLMs Hallucination Mitigation](https://mlsanity.com/paper/2606.07647/) (2026)
- [The Truth Stays in the Family: Enhancing Contextual Grounding via Inherited Truthful Heads in Model Lineages](https://mlsanity.com/paper/2606.15821/) (2026)
- [When Structured Sparse Autoencoders Learn Consistent Concepts Across Modalities](https://mlsanity.com/paper/2607.08605/) (2026)
- [Attention to details, logits to truth: visual-aware attention and logits enhancement to mitigate hallucinations in LVLMs](https://mlsanity.com/paper/2602.09521/) (2026)
- [Cross-modal Identity Mapping: Minimizing Information Loss in Modality Conversion via Reinforcement Learning](https://mlsanity.com/paper/2603.01696/) (2026)
- [HaloProbe: Bayesian Detection and Mitigation of Object Hallucinations in Vision-Language Models](https://mlsanity.com/paper/2604.06165/) (2026)
- [Combating Visual Neglect and Semantic Drift in Large Multimodal Models for Enhanced Cross-Modal Retrieval](https://mlsanity.com/paper/2604.25273/) (2026)
- [Brevity is the Soul of Inference Efficiency: Inducing Concision in VLMs via Data Curation](https://mlsanity.com/paper/2606.25432/) (2026)
- [SAGE: Sink-Aware Guided Emphasis for Visual Grounding in Vision-Language Decoders](https://mlsanity.com/paper/2610.11469/) (2026)
- [From Glance to Focus: Mitigating Recognition Laziness in MLLMs for Fine-Grained Recognition](https://mlsanity.com/paper/2610.13154/) (2026)

## Abstract (arXiv)

Detailed image captioning is essential for tasks like data generation and aiding visually impaired individuals. High-quality captions require a balance between precision and recall, which remains challenging for current multimodal large language models (MLLMs). In this work, we hypothesize that this limitation stems from weakening and increasingly noisy visual attention as responses lengthen. To address this issue, we propose SPARC (Selective Progressive Attention ReCalibration), a training-free method that enhances the contribution of visual tokens during decoding. SPARC is founded on three key observations: (1) increasing the influence of all visual tokens reduces recall; thus, SPARC selectively amplifies visual tokens; (2) as captions lengthen, visual attention becomes noisier, so SPARC identifies critical visual tokens by leveraging attention differences across time steps; (3) as visual attention gradually weakens, SPARC reinforces it to preserve its influence. Our experiments, incorporating both automated and human evaluations, demonstrate that existing methods improve the precision of MLLMs at the cost of recall. In contrast, our proposed method enhances both precision and recall with minimal computational overhead.