Contrastive Language-Image Pre-training (CLIP) has drawn increasing attention recently for its transferable visual representation learning. However, due to the semantic gap within datasets, CLIP's pre-trained image-text alignment becomes sub-optimal on downstream tasks, which severely harms its transferring performance. To better adapt the cross-modality embedding space, we propose to enhance CLIP via Visual-guided Texts, named VT-CLIP. Specifically, we guide textual features of different categories to adaptively explore informative regions on the image and aggregate visual features by attention mechanisms. In this way, the texts become visual-guided, namely, more semantically correlated with downstream images, which greatly benefits the category-wise matching process. In few-shot settings, we evaluate our VT-CLIP on 11 well-known classification datasets to demonstrate its effectiveness.
@article{arxiv.2112.02399,
title = {VT-CLIP: Enhancing Vision-Language Models with Visual-guided Texts},
author = {Longtian Qiu and Renrui Zhang and Ziyu Guo and Ziyao Zeng and Zilu Guo and Yafeng Li and Guangnan Zhang},
journal= {arXiv preprint arXiv:2112.02399},
year = {2023}
}