Words or Vision: Do Vision-Language Models Have Blind Faith in Text?

• URL: http://arxiv.org/abs/2503.02199v1
• Date: Tue, 04 Mar 2025 02:21:07 GMT
• Title: Words or Vision: Do Vision-Language Models Have Blind Faith in Text?
• Authors: Ailin Deng, Tri Cao, Zhirui Chen, Bryan Hooi,
• Abstract summary: Vision-Language Models (VLMs) excel in integrating visual and textual information for vision-centric tasks.<n>We investigate VLMs' modality preferences when faced with visual data and varied textual inputs in vision-centered settings.<n>We discover a emphblind faith in text'' phenomenon:VLMs disproportionately trust textual data over visual data when inconsistencies arise.
• Score: 34.88114876390461
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Vision-Language Models (VLMs) excel in integrating visual and textual information for vision-centric tasks, but their handling of inconsistencies between modalities is underexplored. We investigate VLMs' modality preferences when faced with visual data and varied textual inputs in vision-centered settings. By introducing textual variations to four vision-centric tasks and evaluating ten Vision-Language Models (VLMs), we discover a \emph{``blind faith in text''} phenomenon: VLMs disproportionately trust textual data over visual data when inconsistencies arise, leading to significant performance drops under corrupted text and raising safety concerns. We analyze factors influencing this text bias, including instruction prompts, language model size, text relevance, token order, and the interplay between visual and textual certainty. While certain factors, such as scaling up the language model size, slightly mitigate text bias, others like token order can exacerbate it due to positional biases inherited from language models. To address this issue, we explore supervised fine-tuning with text augmentation and demonstrate its effectiveness in reducing text bias. Additionally, we provide a theoretical analysis suggesting that the blind faith in text phenomenon may stem from an imbalance of pure text and multi-modal data during training. Our findings highlight the need for balanced training and careful consideration of modality interactions in VLMs to enhance their robustness and reliability in handling multi-modal data inconsistencies.

Related papers

• Text2VLM: Adapting Text-Only Datasets to Evaluate Alignment Training in Visual Language Models [0.0]
  Existing evaluation datasets lean towards text-only prompts, leaving visual vulnerabilities under evaluated.<n>We propose Text2VLM, a novel multi-stage pipeline that adapts text-only datasets into multimodal formats.<n>Text2VLM provides a scalable tool for comprehensive safety assessment, contributing to the development of more robust safety mechanisms for Visual Language Models.
  arXiv  Detail & Related papers  (2025-07-28T10:57:44Z)
• Text Speaks Louder than Vision: ASCII Art Reveals Textual Biases in Vision-Language Models [93.46875303598577]
  Vision-language models (VLMs) have advanced rapidly in processing multimodal information, but their ability to reconcile conflicting signals remains underexplored. This work investigates how VLMs process ASCII art, a unique medium where textual elements collectively form visual patterns, potentially creating semantic-visual conflicts.
  arXiv  Detail & Related papers  (2025-04-02T10:47:07Z)
• TextInVision: Text and Prompt Complexity Driven Visual Text Generation Benchmark [61.412934963260724]
  Existing diffusion-based text-to-image models often struggle to accurately embed text within images. We introduce TextInVision, a large-scale, text and prompt complexity driven benchmark to evaluate the ability of diffusion models to integrate visual text into images.
  arXiv  Detail & Related papers  (2025-03-17T21:36:31Z)
• VilBias: A Study of Bias Detection through Linguistic and Visual Cues , presenting Annotation Strategies, Evaluation, and Key Challenges [2.2751168722976587]
  VLBias is a framework that leverages state-of-the-art Large Language Models (LLMs) and Vision-Language Models (VLMs) to detect linguistic and visual biases in news content.<n>We present a multimodal dataset comprising textual content and corresponding images from diverse news sources.
  arXiv  Detail & Related papers  (2024-12-22T15:05:30Z)
• Looking Beyond Text: Reducing Language bias in Large Vision-Language Models via Multimodal Dual-Attention and Soft-Image Guidance [67.26434607115392]
  Large vision-language models (LVLMs) have achieved impressive results in various vision-language tasks. LVLMs suffer from hallucinations caused by language bias, leading to diminished focus on images and ineffective visual comprehension. We propose LACING to address the language bias of LVLMs with muLtimodal duAl-attention meChanIsm (MDA) aNd soft-image Guidance (IFG)
  arXiv  Detail & Related papers  (2024-11-21T16:33:30Z)
• Mitigating Hallucination in Visual-Language Models via Re-Balancing Contrastive Decoding [11.719774461701897]
  We propose Re-Balancing Contrastive Decoding (RBD) method to recalibrate attention distribution in Visual-Language Models. Our method, RBD, outperforms the existing methods by the CHAIR and POPE metrics, mitigate hallucinations without reducing the model's general capabilities.
  arXiv  Detail & Related papers  (2024-09-10T13:13:14Z)
• FINEMATCH: Aspect-based Fine-grained Image and Text Mismatch Detection and Correction [66.98008357232428]
  We propose FineMatch, a new aspect-based fine-grained text and image matching benchmark. FineMatch focuses on text and image mismatch detection and correction. We show that models trained on FineMatch demonstrate enhanced proficiency in detecting fine-grained text and image mismatches.
  arXiv  Detail & Related papers  (2024-04-23T03:42:14Z)
• ConTextual: Evaluating Context-Sensitive Text-Rich Visual Reasoning in Large Multimodal Models [92.60282074937305]
  We introduce ConTextual, a novel dataset featuring human-crafted instructions that require context-sensitive reasoning for text-rich images. We conduct experiments to assess the performance of 14 foundation models and establish a human performance baseline. We observe a significant performance gap of 30.8% between GPT-4V and human performance.
  arXiv  Detail & Related papers  (2024-01-24T09:07:11Z)
• Lost in Translation: When GPT-4V(ision) Can't See Eye to Eye with Text. A Vision-Language-Consistency Analysis of VLLMs and Beyond [7.760124498553333]
  We study whether vision-language models execute vision and language tasks consistently or independently. We introduce a systematic framework that quantifies the capability disparities between different modalities in the multi-modal setting. We introduce "Vision Description Prompting," a method that effectively improves performance in challenging vision-related tasks.
  arXiv  Detail & Related papers  (2023-10-19T06:45:11Z)
• Dynamic Visual Semantic Sub-Embeddings and Fast Re-Ranking [0.5242869847419834]
  We propose a Dynamic Visual Semantic Sub-Embeddings framework (DVSE) to reduce the information entropy. To encourage the generated candidate embeddings to capture various semantic variations, we construct a mixed distribution. We compare the performance with existing set-based method using four image feature encoders and two text feature encoders on three benchmark datasets.
  arXiv  Detail & Related papers  (2023-09-15T04:39:11Z)
• OCRBench: On the Hidden Mystery of OCR in Large Multimodal Models [122.27878464009181]
  We conducted a comprehensive evaluation of Large Multimodal Models, such as GPT4V and Gemini, in various text-related visual tasks. OCRBench contains 29 datasets, making it the most comprehensive OCR evaluation benchmark available.
  arXiv  Detail & Related papers  (2023-05-13T11:28:37Z)
• A Multi-Modal Context Reasoning Approach for Conditional Inference on Joint Textual and Visual Clues [23.743431157431893]
  Conditional inference on joint textual and visual clues is a multi-modal reasoning task. We propose a Multi-modal Context Reasoning approach, named ModCR. We conduct extensive experiments on two corresponding data sets and experimental results show significantly improved performance.
  arXiv  Detail & Related papers  (2023-05-08T08:05:40Z)
• Localization vs. Semantics: Visual Representations in Unimodal and Multimodal Models [57.08925810659545]
  We conduct a comparative analysis of the visual representations in existing vision-and-language models and vision-only models. Our empirical observations suggest that vision-and-language models are better at label prediction tasks. We hope our study sheds light on the role of language in visual learning, and serves as an empirical guide for various pretrained models.
  arXiv  Detail & Related papers  (2022-12-01T05:00:18Z)

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.