We thank all reviewers for their efforts and positive comments

We thank all reviewers for their efforts and positive comments. To Review #1: Thank you for the comments and we will follow the suggestions to do more experiments and analyses. To Review #2: Thank you very much for confirming our contributions. Q1) [Strong priors]: "geometry-consistency" is not a manually designed prior for specific scenes. Our contribution is more deeply leveraging SfM theory in unsupervised depth learning.