Test-time Alignment-Enhanced Adapter for Vision-Language Models

TL;DR

This paper introduces TAEA, a test-time adaptation method for vision-language models that trains an adapter to improve text features during testing, enhancing alignment and performance under distribution shifts.

Contribution

The paper proposes a novel test-time adapter that adjusts text features in VLMs, incorporating negative cache for improved alignment and outperformance of existing methods.

Findings

Outperforms state-of-the-art TTA methods by 0.75% on out-of-distribution benchmarks.

Achieves 2.5% improvement on cross-domain benchmarks.

Maintains acceptable training time for test-time adaptation.

Abstract

Test-time adaptation with pre-trained vision-language models (VLMs) has attracted increasing attention for tackling the issue of distribution shift during the test phase. While prior methods have shown effectiveness in addressing distribution shift by adjusting classification logits, they are not optimal due to keeping text features unchanged. To address this issue, we introduce a new approach called Test-time Alignment-Enhanced Adapter (TAEA), which trains an adapter with test samples to adjust text features during the test phase. We can enhance the text-to-image alignment prediction by utilizing an adapter to adapt text features. Furthermore, we also propose to adopt the negative cache from TDA as enhancement module, which further improves the performance of TAEA. Our approach outperforms the state-of-the-art TTA method of pre-trained VLMs by an average of 0.75% on the out-of-distribution benchmark and 2.5% on the cross-domain benchmark, with an acceptable training time. Code will be available at https://github.com/BaoshunWq/clip-TAEA.