Unifying Visual and Semantic Feature Spaces with Diffusion Models for Enhanced Cross-Modal Alignment

Yuze Zheng, Zixuan Li, Xiangxian Li +4

Image classification models often demonstrate unstable performance in real-world applications due to variations in image information, driven by differing visual perspectives of sub…