Research map / Vision-language and multimodal / Multimodal LLMs
ImpText: A Benchmark and Tool-Augmented Framework for Implicit Text Reasoning
Litao Guo, Jinsong Zhou, Shuaibo Li, Man CHEN, Xinli Xu, Zixin Zhang, Harold Haodong Chen, Ying-Cong Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Introduces ImpText, a benchmark for implicit text (physically deformed, visually camouflaged, or cognitively suggested text) and a tool-augmented framework for reasoning over it in MLLMs.
Problem
MLLMs excel at standard text extraction but struggle with implicit text, where malicious content is concealed via physical deformation, visual camouflage, or cognitive suggestion to evade content moderation.
Approach
A benchmark targeting implicit text concealment techniques, paired with a tool-augmented framework that lets MLLMs invoke tools to recover and reason over hidden text. Based only on the abstract; the specific mechanism is not described.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add tool augmentation alone (ablation) | Qwen3-VL 8B · ImpText-Bench | F1: 26.09 -> 28.62 | No clear effect |
| add knowledge warm-up SFT (ablation) | Qwen3-VL 8B + tools · ImpText-Bench | TMS: 10.63 -> 35.79 | Improved |
| add oracle tool exploration (ablation) | ImpText-Reader · ImpText-Bench | TMS: 35.79 -> 46.83 | Improved |
| add boundary-aware RL (GRPO) (ablation) | ImpText-Reader · ImpText-Bench | F1: 78.49 -> 87.97 | Improved |
| filter training data to 0<P<1 instead of full dataset (ablation) | Semantic Reasoner SFT · ImpText-Bench | TMS: 39.06 -> 46.83 | Improved |
| decouple enhancer and reasoner training instead of joint (ablation) | ImpText-Reader · ImpText-Bench | TMS: 8.38 -> 35.79 | Improved |
| plain GRPO on Qwen3-VL 8B vs base | Qwen3-VL 8B · ImpText-Bench | TMS: 10.22 -> 27.40 | Improved |
| relax NED threshold from 0.5 to 0.9 (ablation) | TMS metric · ImpText-Bench | TMS: 48.26 -> 72.39 (ImpText-Reader) | No clear effect |
| threshold tau=0.5 versus 0.1 or 0.9 for agreement with semantic judge (ablation) | TMS metric · ImpText-Bench 1,000 subset | decision agreement: 0.909 at tau=0.5 vs 0.840 / 0.827 | Improved |
Experimental setup
Benchmarks: ImpText
Closest papers by meaning
- TextShield-R1: Reinforced Reasoning for Tampered Text Detection (AAAI 2026 · similarity 0.86)
- ToxicTextCLIP: Text-Based Poisoning and Backdoor Attacks on CLIP Pre-training (NeurIPS 2025 · similarity 0.86)
- Adversarial Paraphrasing: A Universal Attack for Humanizing AI-Generated Text (NeurIPS 2025 · similarity 0.85)
- ConTextual: Evaluating Context-Sensitive Text-Rich Visual Reasoning in Large Multimodal Models (ICML 2024 · similarity 0.85)
- ImpScore: A Learnable Metric For Quantifying The Implicitness Level of Sentences (ICLR 2025 · similarity 0.85)
- DetectRL: Benchmarking LLM-Generated Text Detection in Real-World Scenarios (NeurIPS 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
