Learning Invariant Feature Points

  • Pascal Fua | École polytechnique fédérale de Lausanne (EPFL)

In this talk, I will present a novel Deep Network architecture that implements the full feature point-handling pipeline, that is, detection, orientation estimation, and feature description. While previous works have successfully tackled each one of these problems, individually, our approach involves learning to do all three in a unified manner while preserving end-to-end differentiability. The resulting pipeline outperforms state-of-the-art methods on a number of benchmark datasets, without having to retrain.

Series: Microsoft Research Talks