Showing papers by "Peter Christiansen published in 2019"

UnsuperPoint: End-to-end Unsupervised Interest Point Detector and Descriptor

Abstract: It is hard to create consistent ground truth data for interest points in natural images, since interest points are hard to define clearly and consistently for a human annotator. This makes interest point detectors non-trivial to build. In this work, we introduce an unsupervised deep learning-based interest point detector and descriptor. Using a self-supervised approach, we utilize a siamese network and a novel loss function that enables interest point scores and positions to be learned automatically. The resulting interest point detector and descriptor is UnsuperPoint. We use regression of point positions to 1) make UnsuperPoint end-to-end trainable and 2) to incorporate non-maximum suppression in the model. Unlike most trainable detectors, it requires no generation of pseudo ground truth points, no structure-from-motion-generated representations and the model is learned from only one round of training. Furthermore, we introduce a novel loss function to regularize network predictions to be uniformly distributed. UnsuperPoint runs in real-time with 323 frames per second (fps) at a resolution of $224\times320$ and 90 fps at $480\times640$. It is comparable or better than state-of-the-art performance when measured for speed, repeatability, localization, matching score and homography estimation on the HPatch dataset.

Direct photon production at low transverse momentum in proton-proton collisions at s =2.76 and 8 TeV

Abstract: Measurements of inclusive and direct photon production at midrapidity in pp collisions at s=2.76 and 8 TeV are presented by the ALICE experiment at the LHC. The results are reported in transverse momentum ranges of pT GeV/c and pT GeV/c, respectively. Photons are detected with the electromagnetic calorimeter (EMCal) and via reconstruction of e+e− pairs from conversions in the ALICE detector material using the central tracking system. For the final measurement of the inclusive photon spectra the results are combined in the overlapping pT interval of both methods. Direct photon spectra, or their upper limits at 90% C.L. are extracted using the direct photon excess ratio Rγ, which quantifies the ratio of inclusive photons over decay photons generated with a decay-photon simulation. An additional hybrid method, combining photons reconstructed from conversions with those identified in the EMCal, is used for the combination of the direct photon excess ratio Rγ, as well as the extraction of direct photon spectra or their upper limits. While no significant signal of direct photons is seen over the full pT range, Rγ for pT>7 GeV/c is at least one σ above unity and consistent with expectations from next-to- leading order pQCD calculations.

Production of the ρ(770)0 meson in pp and Pb-Pb collisions at √sNN = 2.76 TeV

Abstract: The production of the ρ(770)0 meson has been measured at midrapidity (|y|<0.5) in pp and centrality differential Pb-Pb collisions at sNN= 2.76 TeV with the ALICE detector at the Large Hadron Collider. The particles have been reconstructed in the ρ(770)0→π+π− decay channel in the transverse-momentum (pT) range 0.5–11 GeV/c. A centrality-dependent suppression of the ratio of the integrated yields 2ρ(770)0/(π++π−) is observed. The ratio decreases by ∼40% from pp to central Pb-Pb collisions. A study of the pT-differential 2ρ(770)0/(π++π−) ratio reveals that the suppression occurs at low transverse momenta, pT<2 GeV/c. At higher momentum, particle ratios measured in heavy-ion and pp collisions are consistent. The observed suppression is very similar to that previously measured for the K*(892)0/K ratio and is consistent with EPOS3 predictions that may imply that rescattering in the hadronic phase is a dominant mechanism for the observed suppression.

Measurement of two-particle correlations with respect to second- and third-order event planes in Au plus Au collisions at root s(NN)=200 GeV

Abstract: We present measurements of azimuthal correlations of charged hadron pairs in sNN=200 GeV Au+Au collisions for the trigger and associated particle transverse-momentum ranges of 14GeV/c) correlations is suppressed compared with that of correlations measured in p+p collisions. At the lowest associated particle pT(0.5

Internal shear cracking in bulk metal forming

Abstract: This paper presents an uncoupled ductile damage criterion for modelling the opening and propagation of internal shear cracks in bulk metal forming The criterion is built upon the original work on

Measurement of prompt D$^0$, D$^+$, D$^{*+}$, and D$^+_s$ production in p$-$Pb collisions at $\mathbf{\sqrt{{\textit s}_{\rm NN}}~=~5.02~TeV}$

Abstract: The measurement of the production of prompt D$^0$, D$^+$, D$^{*+}$, and D$^+_s$ mesons in proton$-$lead (p$-$Pb) collisions at the centre-of-mass energy per nucleon pair of $\sqrt{s_{\rm NN}}$ = 5.02 TeV, with an integrated luminosity of $292\pm 11$ $\mu$b$^{-1}$, are reported. Differential production cross sections are measured at mid-rapidity ($-0.96

Most scientists prefer small and mid-sized research grants.

Abstract: Scientists’ own perspectives on research funding are often missing. To address this, we surveyed Danish scientists about their ideal research grant. In contrast to a trend towards larger grants, most surveyed scientists prefer small or mid-sized grants to pursue their ideas and advance their scientific careers.

Sparse-to-Dense Depth Completion in Precision Farming

Abstract: Autonomous driving in agriculture can be eased and be more safe if guided by dense depth maps, since dense depth maps outlines scene geometry. RGB monocular image has only naive information about depth and although LiDAR has accurate depth information, it can only provide sparse depth maps. By interpolating sparse LiDAR with aligned color image, reliable dense depth maps can be created. In this paper, we apply a deep regression model where an RGB monocular image was used for a sparse-to-dense LiDAR depth map completion. Our model is based on U-Net architecture presented in [9]. Training the model on the Fieldsafe dataset which is a multi-modal agricultural dataset, however, leads to overfitting. Therefore, we trained the model on the Kitti dataset with high image diversity and test it on the Fieldsafe. We produced an error map to analyze performance of the model for close or far distant objects in the Fieldsafe dataset. The error maps show the absolute difference between the depth ground truth and the predicted depth value. The model preforms 63.6% better on close distance objects than far objects in Fieldsafe. However, the model performs 10.96% better on far objects than close objects in the Kitti dataset.

Friction Compensation in the Compression Test

Abstract: This chapter presents an innovative numerical and experimental methodology for determining the stress-strain curve of metallic materials from the measurements of force and displacement obtained in the upsetting of cylindrical test specimens, which takes into account the influence of friction at the contact interfaces with the die platens. The methodology is based on minimizing the error between the average surface pressure obtained from the experimental measurements of the force and displacement and that obtained from the slab method of analysis of metal plasticity. The computer implementation of the proposed methodology is discussed and a source code written in MATLAB is provided for those readers interested in determining the stress-strain curves of metallic materials. The accuracy and reliability of the methodology and associated computer implementation is assessed by comparing the stress-strain curves obtained from cylindrical and Rastegaev upset test specimens with different lubricants and from compression tests without friction.

A sheet metal necking formability diagram for nonlinear strain paths

Abstract: A new procedure for drawing forming limit curves is suggested. The theoretical basis for computing the forming limit curve due to diffuse necking, for nonlinear strain paths, is derived. The theore...

Revisiting Veerman’s interpolation method

Abstract: This article describes an investigation of Veerman’s interpolation method and its applicability for determining sheet metal formability. The theoretical foundation is established and its mathematical assumptions are clarified. An exact Lagrangian interpolation scheme is also established for comparison. Bulge testing and tensile testing of aluminium sheets containing electro-chemically etched circle grids are performed to experimentally determine the forming limit of the sheet material. The forming limit is determined using (a) Veerman’s interpolation method, (b) exact Lagrangian interpolation and (c) FE-simulations. A comparison of the determined forming limits yields insignificant differences in the limit strain obtained with Veerman’s method or exact Lagrangian interpolation for the two sheet metal forming processes investigated. The agreement with the FE-simulations is reasonable.