Learned Structure from Motion
Develops deep learning approaches to recover camera poses and 3D scene geometry from image matches without requiring initial camera estimates.
2 papers
Where this stands
The written synthesis of this thread is for subscribers. Subscribe.