This chapter focused on SfM and its implementation with OpenCV's sfm contributed module and OpenMVS. We explored some theoretical concepts in multiple view geometry, and several practical matters: extracting key feature points, matching them, creating and analyzing the match graph, running the reconstruction, and finally performing MVS to densify the sparse 3D point cloud.
In the next chapter, we will see how to use OpenCV's face contrib module to detect facial landmarks in photos, as well as detecting the direction a face is pointing with the solvePnP function.
