1 paper
JiYuan Wang, Chunyu Lin, Lei Sun +5
Leveraging visual priors from pre-trained text-to-image (T2I) generative models has shown success in dense prediction. However, dense prediction is inherently an image-to-image tas…