Abstract: Recent contrastive multimodal vision-language models like CLIP have demonstrated robust open-world semantic understanding, becoming the standard image backbones for vision-language ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback