Daniel Cremers

TL;DR: A provably convergent alternating reweighted least-squares scheme for solving the original system of nonlinear PDEs, which are linearized using image ratios in the case of RGB images is introduced.

TL;DR: The concentration of human motion data on lower-dimensional manifolds, approves kernel density estimation as a transparent tool that is able to model priors on arbitrary mixtures of human motions.

TL;DR: This work introduces a self-attention and orientation encoding network (SOE-Net) that fully explores the relationship between points and incorporates long-range context into point-wise local descriptors and proposes a novel loss function called HPHN quadruplet loss, that achieves better performance than the commonly used metric learning loss.

TL;DR: This letter reconstructs a set of non-linear factors that make an optimal approximation of the information on the trajectory accumulated by VIO that make the roll and pitch angles of the global map observable, and improve the robustness and the accuracy of the mapping.

TL;DR: The authors proposed GN-Net, a network optimized with the novel Gauss-Newton loss for training weather invariant deep features, tailored for direct image alignment, which can be trained with pixel correspondences between images taken from different sequences.