Towards Robustness against Typographic Attack with Training-free Concept Localization
Models trained via Contrastive Language-Image Pretraining (CLIP) serve as the foundational vision encoders for most modern Large Vision Language Models (LVLMs). Despite their widespread adoption, CLIP models exhibit a critical yet underexplored failure mode: irrelevant text appearing within images confounds visual representations, biasing them toward lexical meaning rather than true visual semantics. This robustness issue, commonly described as a Typographic Attack (TA), exposes a vulnerability that poses a significant risk to safety-critical applications such as autonomous driving. To achieve
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
- Linked via arxiv authorBohan Liu →
Towards Robustness against Typographic Attack with Training-free Concept Localization
- Linked via arxiv authorWenqian Ye →
Towards Robustness against Typographic Attack with Training-free Concept Localization
- Linked via arxiv authorGuangzhi Xiong →
Towards Robustness against Typographic Attack with Training-free Concept Localization
- Linked via arxiv authorZhenghao He →
Towards Robustness against Typographic Attack with Training-free Concept Localization
- Linked via arxiv authorSanchit Sinha →
Towards Robustness against Typographic Attack with Training-free Concept Localization
- Linked via arxiv authorAidong Zhang →
Towards Robustness against Typographic Attack with Training-free Concept Localization