DocIntent: Answerability-Guided Agentic Restoration for Real-World Document Visual Question Answering
Computer Vision and Pattern RecognitionArtificial Intelligence
Summary
The authors study how various real-world damages like blur and shadows make it hard for multimodal large language models (MLLMs) to answer questions about documents. They propose DocIntent, a method that smartly decides which restoration tools to use based on whether a question can be answered from the image, improving the quality of answers without extra training. DocIntent also uses a rollback system to undo any restoration steps that make answering harder. Their tests show this approach helps models perform better on challenging document images.
Authors
Zihan Huang, Shihang Wu, Junle Liu, Peirong Zhang, Yongxin Shi, Xuhan Zheng, Lianwen Jin
Abstract
Real-world degradations such as blur, shadow, distortion, and moire patterns severely impair the document question-answering capabilities of Multimodal Large Language Models (MLLMs). Applying restoration tools before Visual Question Answering (VQA) is an intuitive solution. However, existing restoration approaches remain limited, as manually designing and executing restoration strategies is labor-intensive and requires domain expertise. Agentic restoration offers new possibilities for automation, yet existing frameworks primarily target natural images and pursue perceptual quality, overlooking that restoration should serve downstream tasks rather than optimize generic image quality metrics. To this end, we explore the value of agentic restoration for real-world degraded document VQA and propose DocIntent, a training-free Answerability-Guided Agentic Restoration framework. DocIntent first assesses question answerability, then identifies task-relevant degradations and selectively invokes restoration tools. A Comparison-Based Rollback mechanism validates each restoration step and reverts it when question-relevant evidence becomes less decipherable. The entire process requires no additional pretrained degradation classifier or image quality assessment model. Extensive experiments on the WildDoc benchmark show that DocIntent consistently improves the average score and consistency of different open- and closed-source MLLMs. The code and experimental data will be publicly available.