Abstract: Recent contrastive multimodal vision-language models like CLIP have demonstrated robust open-world semantic understanding, becoming the standard image backbones for vision-language ...
We propose to rectify Rolling Shutter (RS) distorted images by directly estimating the intermediate distortion flow from the underlying Global Shutter (GS) image to the RS image. This method differs ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results
Feedback