Browsing by Author "Rivinius, Marc"

Now showing 1 - 3 of 3

Open Access
Accountable secure multi-party computation for tally-hiding e-voting
(2020) Rivinius, Marc
With multi-party computation becoming more and more efficient and thus more practical, we can start to investigate application scenarios. One application where multi-party computation can be used to great effect is e-voting. Unlike classical e-voting protocols, one can get tally-hiding e-voting systems. There, some part of the tally (especially the whole set of votes) is not made public. Notwithstanding this, most existing (verifiable) multi-party computation protocols are not suitable for e-voting. A property that is arguably more important than verifiability is missing: accountability -- as a matter of fact, we need external accountability in this setting, where anyone audit the protocol. This is especially of importance for e-voting systems and more researchers are paying attention to it lately. To this effect, we introduce a general multi-party computation protocol that meets all the requirements to be used in e-voting systems. Our protocol achieves accountability and fairness in the honest majority setting and is -- to our best knowledge -- the first one to do so.
Open Access
High-dynamic-range visualization of density plots
(2017) Rivinius, Marc
Density maps are an important means of data representation and have been widely used in various visualizations, e.g., scatter plots, parallel coordinates, and trajectories. Typically, density maps have high dynamic-ranges which are beyond the displayable intensities on a monitor. The common operators to map the data values to displayable intensities (for example, linear, logarithmic, and gamma mappings) do not work in all situations and produce unsatisfactory results, where features may be lost or misleading visualizations may be created. Therefore, we propose a perceptual-based model to better visualize high-dynamic-range density maps: we map high-dynamic-range data to a displayable range through a perceptual tone mapping operator; on top of that, we apply glare simulation to highlight high-density regions which are found by our automatic bright pixel detector. The glare is used to highlight high-density regions, while the tone mapping preserves structural details. In addition, we evaluate different tone mapping operators on density maps in typical data visualizations, which has not been studied to the best of our knowledge. For the whole approach, an efficient GPU-based implementation and an easy-to-use application with intuitive user interactions are provided. We demonstrate the effectiveness of our method through a wide range of density map visualizations.
Open Access
MS-RAFT+ : high resolution multi-scale RAFT
(2023) Jahedi, Azin; Luz, Maximilian; Rivinius, Marc; Mehl, Lukas; Bruhn, Andrés
Hierarchical concepts have proven useful in many classical and learning-based optical flow methods regarding both accuracy and robustness. In this paper we show that such concepts are still useful in the context of recent neural networks that follow RAFT’s paradigm refraining from hierarchical strategies by relying on recurrent updates based on a single-scale all-pairs transform. To this end, we introduce MS-RAFT+: a novel recurrent multi-scale architecture based on RAFT that unifies several successful hierarchical concepts. It employs a coarse-to-fine estimation to enable the use of finer resolutions by useful initializations from coarser scales. Moreover, it relies on RAFT’s correlation pyramid that allows to consider non-local cost information during the matching process. Furthermore, it makes use of advanced multi-scale features that incorporate high-level information from coarser scales. And finally, our method is trained subject to a sample-wise robust multi-scale multi-iteration loss that closely supervises each iteration on each scale, while allowing to discard particularly difficult samples. In combination with an appropriate mixed-dataset training strategy, our method performs favorably. It not only yields highly accurate results on the four major benchmarks (KITTI 2015, MPI Sintel, Middlebury and VIPER), it also allows to achieve these results with a single model and a single parameter setting. Our trained model and code are available at https://github.com/cv-stuttgart/MS_RAFT_plus .