Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Towards Fine-Grained Robustness: Attention-Guided Test-Time Prompt Tuning for Vision-Language Models

About

Vision-Language Models (VLMs), such as CLIP, have achieved significant zero-shot performance on downstream tasks with various fine-tuning adaptation methods. However, recent studies have proven that adversarial attacks can significantly degrade the inference ability of VLMs, posing substantial risks to their practical applications. Prevalent test-time adaptation methods typically rely on multi-view augmentation to implement various fine-tuning strategies, which struggle to identify semantic information and are prone to destroying discriminative regions in fine-grained scenarios. To address these limitations, we propose Attention-Guided Test-Time Prompt Tuning (A-TPT), a semantics-preserving method designed for test-time adaptation. We first refine the gradient attention rollout mechanism to identify semantically meaningful regions surviving under adversarial attacks. Furthermore, we leverage them to guide the spatially varying augmentation intensities and multi-view ensemble for prompt tuning and inference. Extensive experiments demonstrate that A-TPT outperforms existing test-time adaptation methods on both adversarial and clean data. Codes are available at https://github.com/SEU-VIPGroup/A-TPT .

Jia-Wei Hai, Yijun Wang, Xiu-Shen Wei• 2026

Related benchmarks

TaskDatasetResultRank
Image ClassificationImageNet-R--
36
Fine-grained Image Classification8 Fine-grained Dataset Suite Average--
35
Image ClassificationStanford Cars--
29
ClassificationCaltech101
Robust Accuracy85.6
22
Fine grained classificationFine-grained classification datasets Average of Caltech101, Pets, Cars, Flower102, Aircraft, DTD, EuroSAT, UCF101 (test)--
20
Image ClassificationImageNet A
Top-1 Accuracy (Clean)31.6
16
Image ClassificationImageNet V2
Top-1 Accuracy (Clean)55.1
16
ClassificationOxford-IIIT Pets
Adversarial Accuracy70.5
12
ClassificationUCF101
Adversarial Accuracy51.7
12
ClassificationEuroSAT
Adversarial Accuracy13.1
12
Showing 10 of 16 rows

Other info

Follow for update