Different from these approaches, our proposed depth-guided vision transformer with a normalizing flows (NF-DVT) network uses normalizing flows to build priors in depth maps to achieve more accurate ...