Align2Ground: Weakly Supervised Phrase Grounding Guided by Image-Caption Alignment.
IEEE International Conference on Computer Vision（2019）
关键词
weakly supervised phrase grounding,matched RoI,align2ground,downstream caption-to-image retrieval task,qualitative analysis,empirical analysis,visual grounding,image encoding,learned representation,subsequent step,discriminative image representation,latent correspondences,phrase localization,downstream task,end-to-end model,image-caption pairs,free-form textual phrases
AI 理解论文
溯源树
样例
生成溯源树，研究论文发展脉络
Chat Paper
正在生成论文摘要