DA/MA/BA Thesis: Learning a Keypoint Detector for Matching Images

In this computer vision project, we aim to learn the best keypoint detector for a specific descriptor, to be used to create correspondences between pairs of images, following the methodology in [Salti15]. Differently from most detectors in literature, handcrafted to maximize a specific image saliency or characteristic (e.g., edges, corners, blobs), in this work we aim to learn the saliency of the detector that best suits the characteristic of a specific descriptor paired to it. The work will be carried out in collaboration with the Computer Vision Lab of the University of Bologna.

Requirements

The student should have experience in C++ and/or Matlab. Knowledge in machine learning, in particular random forest, is also helpful.

References

[Salti15] S. Salti, F. Tombari, R. Spezialetti, L. Di Stefano, Learning a Descriptor-Specific 3D Keypoint Detector (ICCV 2015)

Contact

Mail to Federico Tombari, or drop by room 03.13.042