1 paper
Tianchen Zhao, Xuanbai Chen, Zhihua Li +5
Recent generative data augmentation methods conditioned on both image and text prompts struggle to balance between fidelity and diversity, as it is challenging to preserve essentia…