# SAFE-G: Structure-aware Faithful Evidence-guided Generation for Knowledge-based Visual Question Answering

arXiv: https://arxiv.org/abs/2608.21796 · HTML: https://mlsanity.com/paper/2608.21796/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 21 of its 44 references that are in the map.
- First posted on arXiv 22 Aug 2026.

## Builds on (most cited first, 21 of 21)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuning](https://mlsanity.com/paper/2305.06500/) (2023)
- [FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning](https://mlsanity.com/paper/2307.08691/) (2023)
- [Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answering](https://mlsanity.com/paper/1612.00837/) (2016)
- [VQA: Visual Question Answering](https://mlsanity.com/paper/1505.00468/) (2015)
- [Billion-scale similarity search with GPUs](https://mlsanity.com/paper/1702.08734/) (2017)
- [Search-R1: Training LLMs to Reason and Leverage Search Engines with Reinforcement Learning](https://mlsanity.com/paper/2503.09516/) (2025)
- [A-OKVQA: A Benchmark for Visual Question Answering using World Knowledge](https://mlsanity.com/paper/2206.01718/) (2022)
- [SFT Memorizes, RL Generalizes: A Comparative Study of Foundation Model Post-training](https://mlsanity.com/paper/2501.17161/) (2025)
- [ZeRO-Infinity: Breaking the GPU Memory Wall for Extreme Scale Deep Learning](https://mlsanity.com/paper/2104.07857/) (2021)
- [Can Pre-trained Vision and Language Models Answer Visual Information-Seeking Questions?](https://mlsanity.com/paper/2302.11713/) (2023)
- [FVQA: Fact-based Visual Question Answering](https://mlsanity.com/paper/1606.05433/) (2016)
- [Visual Question Answering: A Survey of Methods and Datasets](https://mlsanity.com/paper/1607.05910/) (2016)
- [Retrieval Augmented Visual Question Answering with Outside Knowledge](https://mlsanity.com/paper/2210.03809/) (2022)
- [EchoSight: Advancing Visual-Language Models with Wiki Knowledge](https://mlsanity.com/paper/2407.12735/) (2024)
- [PreFLMR: Scaling Up Fine-Grained Late-Interaction Multi-modal Retrievers](https://mlsanity.com/paper/2402.08327/) (2024)
- [Augmenting Multimodal LLMs with Self-Reflective Tokens for Knowledge-based Visual Question Answering](https://mlsanity.com/paper/2411.16863/) (2024)
- [MMKB-RAG: A Multi-Modal Knowledge-Based Retrieval-Augmented Generation Framework](https://mlsanity.com/paper/2504.10074/) (2025)

## Abstract (arXiv)

Knowledge-based Visual Question Answering (KB-VQA) aims to answer queries that necessitate reasoning over external knowledge sources beyond the visual content. Typically, current methods fuse multimodal features to retrieve external information, subsequently leveraging Multimodal Large Language Models (MLLMs) to derive answers from the retrieved evidence. However, these methods often struggle to capture structural associations within complex contexts to effectively filter noise. Furthermore, they frequently fail to ensure that the reasoning process remains strictly faithful to the retrieved evidence. To address these challenges, we propose SAFE-G, a Structure-Aware Faithful Evidence-guided Generation framework, which enables precise evidence localization and trustworthy reasoning. Specifically, we first employ a coarse-grained hybrid search fusing visual and textual modalities to recall candidate documents, and subsequently implement a structure-aware fine-grained graph retrieval that captures structural dependencies to filter noise and pinpoint precise evidence. Moreover, we introduce a reinforcement learning (RL) strategy with an evidence-grounded reward that assigns credit to correct answers only when the selected evidence is correct. This strict alignment constraint compels the model to anchor its response in the retrieved context, effectively enhancing its capability to locate evidence via multimodal features and perform faithful reasoning. Extensive experiments on the Encyclopedic-VQA and InfoSeek benchmarks demonstrate that SAFE-G outperforms prior methods by a margin of 8.9% and 3.5%, substantially enhancing the overall reasoning accuracy. Our source code is publicly available at: https://github.com/MINE-USTC/SAFE-G.