@unpublished{Xu:2020, author = {Xu, B and Davison, AJ and Leutenegger, S}, publisher = {arXiv}, title = {Deep probabilistic feature-metric tracking}, url = {http://arxiv.org/abs/2008.13504v1}, year = {2020} }
TY - UNPB AB - Dense image alignment from RGB-D images remains a critical issue forreal-world applications, especially under challenging lighting conditions andin a wide baseline setting. In this paper, we propose a new framework to learna pixel-wise deep feature map and a deep feature-metric uncertainty mappredicted by a Convolutional Neural Network (CNN), which together formulate adeep probabilistic feature-metric residual of the two-view constraint that canbe minimised using Gauss-Newton in a coarse-to-fine optimisation framework.Furthermore, our network predicts a deep initial pose for faster and morereliable convergence. The optimisation steps are differentiable and unrolled totrain in an end-to-end fashion. Due to its probabilistic essence, our approachcan easily couple with other residuals, where we show a combination with ICP.Experimental results demonstrate state-of-the-art performance on the TUM RGB-Ddataset and 3D rigid object tracking dataset. We further demonstrate ourmethod's robustness and convergence qualitatively. AU - Xu,B AU - Davison,AJ AU - Leutenegger,S PB - arXiv PY - 2020/// TI - Deep probabilistic feature-metric tracking UR - http://arxiv.org/abs/2008.13504v1 UR - http://hdl.handle.net/10044/1/82605 ER -