Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models

Jinhao Li, Haopeng Li, Sarah Monazam Erfani, Lei Feng, James Bailey, Feng Liu
Proceedings of the 41st International Conference on Machine Learning, PMLR 235:28018-28039, 2024.

Abstract

It has recently been discovered that using a pre-trained vision-language model (VLM), e.g., CLIP, to align a whole query image with several finer text descriptions generated by a large language model can significantly enhance zero-shot performance. However, in this paper, we empirically find that the finer descriptions tend to align more effectively with local areas of the query image rather than the whole image, and then we theoretically validate this finding. Thus, we present a method called weighted visual-text cross alignment (WCA). This method begins with a localized visual prompting technique, designed to identify local visual areas within the query image. The local visual areas are then cross-aligned with the finer descriptions by creating a similarity matrix using the pre-trained VLM. To determine how well a query image aligns with each category, we develop a score function based on the weighted similarities in this matrix. Extensive experiments demonstrate that our method significantly improves zero-shot performance across various datasets, achieving results that are even comparable to few-shot learning methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v235-li24ag, title = {Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models}, author = {Li, Jinhao and Li, Haopeng and Erfani, Sarah Monazam and Feng, Lei and Bailey, James and Liu, Feng}, booktitle = {Proceedings of the 41st International Conference on Machine Learning}, pages = {28018--28039}, year = {2024}, editor = {Salakhutdinov, Ruslan and Kolter, Zico and Heller, Katherine and Weller, Adrian and Oliver, Nuria and Scarlett, Jonathan and Berkenkamp, Felix}, volume = {235}, series = {Proceedings of Machine Learning Research}, month = {21--27 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v235/main/assets/li24ag/li24ag.pdf}, url = {https://proceedings.mlr.press/v235/li24ag.html}, abstract = {It has recently been discovered that using a pre-trained vision-language model (VLM), e.g., CLIP, to align a whole query image with several finer text descriptions generated by a large language model can significantly enhance zero-shot performance. However, in this paper, we empirically find that the finer descriptions tend to align more effectively with local areas of the query image rather than the whole image, and then we theoretically validate this finding. Thus, we present a method called weighted visual-text cross alignment (WCA). This method begins with a localized visual prompting technique, designed to identify local visual areas within the query image. The local visual areas are then cross-aligned with the finer descriptions by creating a similarity matrix using the pre-trained VLM. To determine how well a query image aligns with each category, we develop a score function based on the weighted similarities in this matrix. Extensive experiments demonstrate that our method significantly improves zero-shot performance across various datasets, achieving results that are even comparable to few-shot learning methods.} }
Endnote
%0 Conference Paper %T Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models %A Jinhao Li %A Haopeng Li %A Sarah Monazam Erfani %A Lei Feng %A James Bailey %A Feng Liu %B Proceedings of the 41st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Ruslan Salakhutdinov %E Zico Kolter %E Katherine Heller %E Adrian Weller %E Nuria Oliver %E Jonathan Scarlett %E Felix Berkenkamp %F pmlr-v235-li24ag %I PMLR %P 28018--28039 %U https://proceedings.mlr.press/v235/li24ag.html %V 235 %X It has recently been discovered that using a pre-trained vision-language model (VLM), e.g., CLIP, to align a whole query image with several finer text descriptions generated by a large language model can significantly enhance zero-shot performance. However, in this paper, we empirically find that the finer descriptions tend to align more effectively with local areas of the query image rather than the whole image, and then we theoretically validate this finding. Thus, we present a method called weighted visual-text cross alignment (WCA). This method begins with a localized visual prompting technique, designed to identify local visual areas within the query image. The local visual areas are then cross-aligned with the finer descriptions by creating a similarity matrix using the pre-trained VLM. To determine how well a query image aligns with each category, we develop a score function based on the weighted similarities in this matrix. Extensive experiments demonstrate that our method significantly improves zero-shot performance across various datasets, achieving results that are even comparable to few-shot learning methods.
APA
Li, J., Li, H., Erfani, S.M., Feng, L., Bailey, J. & Liu, F.. (2024). Visual-Text Cross Alignment: Refining the Similarity Score in Vision-Language Models. Proceedings of the 41st International Conference on Machine Learning, in Proceedings of Machine Learning Research 235:28018-28039 Available from https://proceedings.mlr.press/v235/li24ag.html.

Related Material