Scale recovery problem

Thank you for your excellent work! I noticed that the camera pose and depth generated by the model are generated within a relative space, rather than the real world. How can the predicted spatial point cloud be reconstructed to fit within the real world? Looking forward to your reply.