让视觉语言模型看懂错觉，通过图像处理和提示设计提升判断准确性。
Illusion-Aware Visual Preprocessing and Anti-Illusion Prompting for Classic Illusion Understanding in Vision-Language Models
- 用特定图像变换削弱错觉干扰，如提取边缘、隔离颜色。
- 在630张测试图上达到90.48%准确率，人类验证子集达98.41%。
- 无需微调，仅靠图像预处理和提示设计，适合部署在现成模型上。
视觉语言模型（VLMs）对视觉错觉存在系统性偏差，倾向于回忆记忆中的事实而非感知真实的视觉差异。本文针对CVPR 2026年DataCV挑战赛第1任务，提出一种无需训练的框架，通过三种互补策略解决感知与记忆的冲突：(1) 错觉感知图像预处理，通过类型特定的变换（边缘提取、颜色隔离、形态学处理、参考线叠加）弱化诱发错觉的上下文；(2) 反错觉提示工程，引导VLM进行定性视觉比较；(3) 多投票集成，进一步提升鲁棒性。该方法在官方630张图像测试集上使用Claude (claude-opus-4-6) 配合5票多数表决，取得90.48%准确率，在人工验证子集上达98.41%。方案不依赖微调，仅依靠视觉操作与提示设计。最终在挑战中获得第二名，仅落后第一名0.47%。代码已公开于https://github.com/jasminezz/sf-illusion-aware-vlm.git。
原文摘要 · Abstract (English)
Vision-Language Models (VLMs) exhibit systematic bias toward visual illusions, recalling memorized facts rather than perceiving actual visual differences. This paper presents a training-free framework for the 5th DataCV Challenge Task 1 at CVPR 2026, addressing this perception-versus-memory conflict through three complementary strategies:(1) illusion-aware image preprocessing that weakens illusion-inducing context via type-specific transformations (edge extraction, color isolation, morphological processing, and reference-line overlay), (2) anti-illusion prompt engineering guiding VLMs toward qualitative visual comparison, and (3) multi-vote ensemble that further improves robustness. Our method achieves 90.48% accuracy on the official 630-image test set using Claude (claude-opus-4-6) with 5-vote majority ensemble, and 98.41% on a human-verified subset. The approach requires no finetuning, relying solely on visual manipulation and prompt design. Our solution secured 2nd place in the challenge, only 0.47% behind the 1st-place solution. Code is available at https://github.com/jasminezz/sf-illusion-aware-vlm.git.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。