no code implementations • 1 Dec 2021 • Boitumelo Ruf, Martin Weinmann, Stefan Hinz
With FaSS-MVS, we present an approach for fast multi-view stereo with surface-aware Semi-Global Matching that allows for rapid depth and normal map estimation from monocular aerial video data captured by UAVs.
no code implementations • 15 Jun 2021 • Boitumelo Ruf, Jonas Mohrs, Martin Weinmann, Stefan Hinz, Jürgen Beyerer
In this, we propose an optimization of the algorithm for embedded CUDA GPUs, by using massively parallel computing, as well as using the NEON intrinsics to optimize the algorithm for vectorized SIMD processing on embedded ARM CPUs.
no code implementations • 21 Apr 2021 • Max Hermann, Boitumelo Ruf, Martin Weinmann
To create a 3D model of the scene, we rely on a three-stage processing chain.
1 code implementation • 17 Aug 2020 • Max Hermann, Boitumelo Ruf, Martin Weinmann, Stefan Hinz
Therefore, in this paper, we present a method for self-supervised learning for monocular depth estimation from aerial imagery that does not require annotated training data.
no code implementations • 21 Sep 2019 • Boitumelo Ruf, Thomas Pollok, Martin Weinmann
One key aspect in this is the efficient dense image matching and depth estimation.
no code implementations • 26 Jun 2019 • Sylvia Schmitz, Martin Weinmann, Boitumelo Ruf
These correspondences are then used in an iterative optimization scheme to refine the initial camera pose by minimizing the reprojection error.
no code implementations • 17 Jul 2018 • Boitumelo Ruf, Sebastian Monka, Matthias Kollmann, Michael Grinberg
Obstacle avoidance is based on a reactive approach which finds the shortest path around the obstacles as soon as they have a critical distance to the UAV.
no code implementations • 23 Apr 2018 • Boitumelo Ruf, Laurenz Thiel, Martin Weinmann
We use transfer learning to train the Faster R-CNN method for real-time deep object detection, by combining a large ground-based dataset for urban scene understanding with a smaller number of images from an aerial dataset.