Filetype
Monocular depth estimation is a highly challenging problem that is often addressed with deep neural networks. While these use recognition of high level image features to predict reasonably looking depth maps,the result often has poor metric accuracy. Moreover,the standard feed forward architecture does not allow modification of the prediction based on cues other than the image.In this paper we rel
Activation functions play a central role in deep learning since they form an essential building stone of neural networks. In the last few years, the focus has been shifting towards investigating new types of activations that outperform the classical Rectified Linear Unit (ReLU) in modern neural architectures. Most recently, rational activation functions (RAFs) have awakened interest because they w
Many hand-held or mixed reality devices are used with a single sensor for 3D reconstruction, although they often comprise multiple sensors. Multi-sensor depth fusion is able to substantially improve the robustness and accuracy of 3D reconstruction methods, but existing techniques are not robust enough to handle sensors which operate with diverse value ranges as well as noise and outlier statistics
There exist several datasets for developing self-driving car methodologies. Manually collected datasets impose inherent limitations on the variability of test cases and it is particularly difficult to acquire challenging scenarios, e.g. ones involving collisions with pedestrians. A way to alleviate this is to consider automatic generation of safety-critical scenarios for autonomous vehicle (AV) te
The phase gradient autofocus (PGA) is a robust autofocusing approach that can efficiently refocus defocused synthetic aperture radar (SAR) imagery produced by frequency-domain algorithms. However, from a conventional viewpoint, PGA cannot be extended to refocus SAR imagery produced by time-domain algorithms, such as the filtered backprojection (FBP), as the spectrum of the FBP imagery is range amb
Impulsive noise is one of the most challenging forms of interference in an underwater acoustic environment. In this paper, we present an underwater acoustic channel deconvolution method based on a sparse representation framework. The application of the method enables a channel impulse response reconstruction that is robust to impulsive noise. By exploiting the inherent structure in the channel res
Measured aerosol size distributions from three measurement stations and modeled air mass trajectory data were combined to study aerosol dynamics in the boreal forest zone in Northern Scandinavia. Three approaches were used: investigation of new particle formation events, analysis of air masses arriving from ocean to continent, and study of changes in the aerosol size distributions when air masses
In compressed sensing a sparse vector is approximately retrieved from an underdetermined equation system Ax = b. Exact retrievalwouldmean solving a large combinatorial problem which is well known to be NP-hard. For b of the form Ax0 + ϵ, where x0 is the ground truth and ϵ is noise, the 'oracle solution' is the one you get if you a priori know the support of x0, and is the best solution one could h
We consider a perturbation problem for embedded eigenvalues of a self-adjoint differential operator in L2(R;Rn). In particular, we study the set of all small perturbations in an appropriate Banach space for which the embedded eigenvalue remains embedded in the continuous spectrum. We show that this set of small perturbations forms a smooth manifold and we specify its co-dimension. Our methods invo
Recently there has been an increase in the number of available autonomous vehicle (AV) models. To evaluate and compare the safety of the various models the AVs need to be tested in several diverse safety-critical scenarios. We propose the Adversarial Test Case Generator (ATCG) that differently from previous test case generators allows for the generation of realistic collision scenarios with varied
It is difficult to perform 3D reconstruction from on-vehicle gathered video due to the large forward motion of the vehicle. Even object detection and human sensing models perform significantly worse on onboard videos when compared to standard benchmarks because objects often appear far away from the camera compared to the standard object detection benchmarks, image quality is often decreased by mo
Realistic modelling of pedestrians in Autonomous Vehicles (AV)s and AV testing is crucial to avoid lethal collisions in deployment. The majority of AV trajectory forecasting literature do not utilize the motion cues present in 3D human pose because it is hard to gather large datasets of articulated 3D pedestrian motion. In this thesis we discuss the difficulties in data gathering and propose a ped
In this paper, we describe the conversion of a heart trans- plantation dataset in the RDF format and its application in a survival prediction program. The International Society for Heart & Lung Trans- plantation (ISHLT) maintains a registry of heart transplantations that it gathers from operations performed worldwide. The US United Net- work for Organ Sharing (UNOS) and the Scandinavian Scandi
In this paper, we propose a novel method for estimating the locations of near- and/or far-field harmonic audio sources impinging on an arbitrary, but calibrated, sensor array. Using a joint pitch and location estimation formed in two steps, we first estimate the fundamental frequencies and complex amplitudes under a sinusoidal model assumption, whereafter the location of each source is found by ut
UPPSATSEN ÄR SKRIVEN I ÄMNET TJECKISKA (KURS TJE252). The history of gender equality in the Czech Republic i's long and diverse. Thi's thesi's aim's to provide a concise mapping of it's development, with the purpose to illustrate and explain factor's that have consequence's to thi's day. The principal focu's of the study i's the civic women''s movem
Background: Patients with diabetes admitted to emergency care face a higher risk of complications, including prolonged hospital stays, admissions to the intensive care unit and mortality. Aim: To develop a machine learning (ML) model to predict 30-day mortality in patients with diabetes admitted to the emergency department (ED). Design and setting: A cohort study utilizing data from all nine ED’s
Biogenic volatile organic compounds (BVOCs) can be released from soils to the atmosphere through microbial decomposition of plant residues or soil organic carbon, root emission, evaporation of litter-stored BVOCs, and other physical processes. Soils can also act as a sink of BVOCs through biotic and abiotic uptake. Currently, the source and sink capabilities of soils have not been explicitly accou
Background: Targeted temperature management is recommended after out-of-hospital cardiac arrest and may be achieved using a variety of cooling devices. This study was conducted to explore the performance and outcomes for intravascular versus surface devices for targeted temperature management after out-of-hospital cardiac arrest. Method: A retrospective analysis of data from the Targeted Temperatu
This paper presents a robust phase-based positioningframework using a massive multiple-input multiple-output(MIMO) system. The phase-based distance estimates of MPCstogether with other parameters are tracked with an ExtendedKalman Filter (EKF), the state dimension of which varies withthe birth-death processes of paths. The iterative maximumlikelihoodestimation algorithm (RIMAX) and the modeling of
