Comment by yubin

Both VisualSfM and OpenSfM use similar methodology (incremental structure from motion) to reconstruct 3D scene points and camera positions from photos.

And yes, you can build your own tools (even commercially) with OpenSfM.