How do you combine solutions from classical geometric methods with deep learning methods in a principled way? We present a framework that learns relative camera pose estimation along with its probabilistic fusion using estimation from geometric methods. The fusion is achieved by learning the network uncertainty under explicit guidance from the geometric uncertainty, thereby learning to take into account the geometric solution in relation to the network prediction. The learning is driven by a self-attention graph neural network.

Collaborators: Manmohan Chandraker

MATERIAL LINKS

PDF
Supplementary