Open-Set Image Tagging with Multi-Grained Text Supervision
This work addresses the challenge of recognizing diverse and unseen categories in image tagging for computer vision applications, representing a strong specific gain rather than a foundational breakthrough.
The paper tackles the problem of open-set image tagging by introducing RAM++, which integrates multi-grained text supervision to improve recognition of both predefined and open-set categories, achieving significant performance gains such as 10.2 mAP and 15.4 mAP improvements over CLIP on OpenImages and ImageNet for predefined tags.
In this paper, we introduce the Recognize Anything Plus Model (RAM++), an open-set image tagging model effectively leveraging multi-grained text supervision. Previous approaches (e.g., CLIP) primarily utilize global text supervision paired with images, leading to sub-optimal performance in recognizing multiple individual semantic tags. In contrast, RAM++ seamlessly integrates individual tag supervision with global text supervision, all within a unified alignment framework. This integration not only ensures efficient recognition of predefined tag categories, but also enhances generalization capabilities for diverse open-set categories. Furthermore, RAM++ employs large language models (LLMs) to convert semantically constrained tag supervision into more expansive tag description supervision, thereby enriching the scope of open-set visual description concepts. Comprehensive evaluations on various image recognition benchmarks demonstrate RAM++ exceeds existing state-of-the-art (SOTA) open-set image tagging models on most aspects. Specifically, for predefined commonly used tag categories, RAM++ showcases 10.2 mAP and 15.4 mAP enhancements over CLIP on OpenImages and ImageNet. For open-set categories beyond predefined, RAM++ records improvements of 5.0 mAP and 6.4 mAP over CLIP and RAM respectively on OpenImages. For diverse human-object interaction phrases, RAM++ achieves 7.8 mAP and 4.7 mAP improvements on the HICO benchmark. Code, datasets and pre-trained models are available at \url{https://github.com/xinyu1205/recognize-anything}.