Tag2Text: Guiding Vision-Language Model via Image Tagging
About
This paper presents Tag2Text, a vision language pre-training (VLP) framework, which introduces image tagging into vision-language models to guide the learning of visual-linguistic features. In contrast to prior works which utilize object tags either manually labeled or automatically detected with an off-the-shelf detector with limited performance, our approach explicitly learns an image tagger using tags parsed from image-paired text and thus provides a strong semantic guidance to vision-language models. In this way, Tag2Text can utilize large-scale annotation-free image tags in accordance with image-text pairs, and provides more diverse tag categories beyond objects. As a result, Tag2Text demonstrates the ability of a foundational image tagging model, with superior zero-shot performance even comparable to fully supervised models. Moreover, by leveraging the tagging guidance, Tag2Text effectively enhances the performance of vision-language models on both generation-based and alignment-based tasks. Across a wide range of downstream benchmarks, Tag2Text achieves state-of-the-art results with similar model sizes and data scales, demonstrating the efficacy of the proposed tagging guidance. Code, demo and pre-trained models are available at https://github.com/xinyu1205/recognize-anything.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Open vocabulary image classification | Wiki-H | Used Objects397 | 10 | |
| Image Classification | World-H (test) | Used Objects176 | 10 | |
| Medical image captioning | RAD | Final Score0.3332 | 4 | |
| Medical image captioning | Slake | Final Score0.2956 | 4 | |
| Human Alignment Correlation | Video-Bench | -- | 3 |