arXiv subjects

Athanasios Papaioannou

Publications and source records attributed to Athanasios Papaioannou.

25 records · Page 2Linked to original sources

arXiv2019

3DFaceGAN: Adversarial Nets for 3D Face Representation, Generation, and Translation

Over the past few years, Generative Adversarial Networks (GANs) have garnered increased interest among researchers in Computer Vision, with applications including, but not limited to, image generation, translation, imputation, and super-resolution. Nevertheless, no GAN-based method has been proposed in the literature that can successfully represent, generate or translate 3D facial shapes (meshes). This can be primarily attributed to two facts, namely that (a) publicly available 3D face databases are scarce as well as limited in terms of sample size and variability (e.g., few subjects, little diversity in race and gender), and (b) mesh convolutions for deep networks present several challenges that are not entirely tackled in the literature, leading to operator approximations and model instability, often failing to preserve high-frequency components of the distribution. As a result, linear methods such as Principal Component Analysis (PCA) have been mainly utilized towards 3D shape analysis, despite being unable to capture non-linearities and high frequency details of the 3D face - such as eyelid and lip variations. In this work, we present 3DFaceGAN, the first GAN tailored towards modeling the distribution of 3D facial surfaces, while retaining the high frequency details of 3D face shapes. We conduct an extensive series of both qualitative and quantitative experiments, where the merits of 3DFaceGAN are clearly demonstrated against other, state-of-the-art methods in tasks such as 3D shape representation, generation, and translation.

cs.CV↗

arXiv2019

Synthesising 3D Facial Motion from "In-the-Wild" Speech

Synthesising 3D facial motion from speech is a crucial problem manifesting in a multitude of applications such as computer games and movies. Recently proposed methods tackle this problem in controlled conditions of speech. In this paper, we introduce the first methodology for 3D facial motion synthesis from speech captured in arbitrary recording conditions ("in-the-wild") and independent of the speaker. For our purposes, we captured 4D sequences of people uttering 500 words, contained in the Lip Reading Words (LRW) a publicly available large-scale in-the-wild dataset, and built a set of 3D blendshapes appropriate for speech. We correlate the 3D shape parameters of the speech blendshapes to the LRW audio samples by means of a novel time-warping technique, named Deep Canonical Attentional Warping (DCAW), that can simultaneously learn hierarchical non-linear representations and a warping path in an end-to-end manner. We thoroughly evaluate our proposed methods, and show the ability of a deep learning model to synthesise 3D facial motion in handling different speakers and continuous speech signals in uncontrolled conditions.

cs.CV↗

arXiv2019

Sources of Solar Energetic Particles

Solar Energetic Particles (SEP) are an integral part of the physical processes related with Space Weather. We present a review for the acceleration mechanisms related to the explosive phenomena (flares and/or CMEs) inside the solar corona. For more than 40 years, the main 2D cartoon representing our understanding of the explosive phenomena inside the solar corona remained almost unchanged. The acceleration mechanisms related to solar flares and CMEs also remained unchanged and were part of the same cartoon. In this review, we revise the standard cartoon and present evidence from recent global MHD simulations that supports the argument that explosive phenomena will lead to the spontaneous formation of current sheets in different parts of the erupting magnetic structure. The evolution of the large scale current sheets and their fragmentation will lead to strong turbulence and turbulent reconnection during solar flares and turbulent shocks. In other words, the acceleration mechanism in flares and CME-driven shocks may be the same, and their difference will be the overall magnetic topology, the ambient plasma parameters, and the duration of the unstable driver.

astro-ph.HE↗

arXiv2019

Deep Affect Prediction in-the-wild: Aff-Wild Database and Challenge, Deep Architectures, and Beyond

Automatic understanding of human affect using visual signals is of great importance in everyday human-machine interactions. Appraising human emotional states, behaviors and reactions displayed in real-world settings, can be accomplished using latent continuous dimensions (e.g., the circumplex model of affect). Valence (i.e., how positive or negative is an emotion) & arousal (i.e., power of the activation of the emotion) constitute popular and effective affect representations. Nevertheless, the majority of collected datasets this far, although containing naturalistic emotional states, have been captured in highly controlled recording conditions. In this paper, we introduce the Aff-Wild benchmark for training and evaluating affect recognition algorithms. We also report on the results of the First Affect-in-the-wild Challenge that was organized in conjunction with CVPR 2017 on the Aff-Wild database and was the first ever challenge on the estimation of valence and arousal in-the-wild. Furthermore, we design and extensively train an end-to-end deep neural architecture which performs prediction of continuous emotion dimensions based on visual cues. The proposed deep learning architecture, AffWildNet, includes convolutional & recurrent neural network layers, exploiting the invariant properties of convolutional features, while also modeling temporal dynamics that arise in human behavior via the recurrent layers. The AffWildNet produced state-of-the-art results on the Aff-Wild Challenge. We then exploit the AffWild database for learning features, which can be used as priors for achieving best performances both for dimensional, as well as categorical emotion recognition, using the RECOLA, AFEW-VA and EmotiW datasets, compared to all other methods designed for the same goal. The database and emotion recognition models are available at http://ibug.doc.ic.ac.uk/resources/first-affect-wild-challenge.

cs.CV↗

arXiv2018

Catalogue of >55 MeV Wide-longitude Solar Proton Events Observed by SOHO, ACE, and the STEREOs at $\approx$1 AU during 2009-2016

Based on energetic particle observations made at $\approx$1 AU, we present a catalogue of 46 wide-longitude (>45°) solar energetic particle (SEP) events detected at multiple locations during 2009-2016. The particle kinetic energies of interest were chosen as >55 MeV for protons and 0.18-0.31 MeV for electrons. We make use of proton data from the Solar and Heliospheric Observatory/Energetic and Relativistic Nuclei and Electron experiment (SOHO/ERNE) and the Solar Terrestrial Relations Observatory/High Energy Telescopes (STEREO/HET), together with electron data from the Advanced Composition Explorer/Electron, Proton, and Alpha Monitor (ACE/EPAM) and the STEREO/Solar Electron and Proton Telescopes (SEPT). We consider soft X-ray data from the Geostationary Operational Environmental Satellites (GOES) and coronal mass ejection (CME) observations made with the SOHO/Large Angle and Spectrometric Coronagraph (LASCO) and STEREO/Coronagraphs 1 and 2 (COR1, COR2) to establish the probable associations between SEP events and the related solar phenomena. Event onset times and peak intensities are determined; velocity dispersion analysis (VDA) and time-shifting analysis (TSA) are performed for protons; TSA is performed for electrons. In our event sample, there is a tendency for the highest peak intensities to occur when the observer is magnetically connected to solar regions west of the flare. Our estimates for the mean event width, derived as the standard deviation of a Gaussian curve modelling the SEP intensities (protons $\approx$44°, electrons $\approx$50°), largely agree with previous results for lower-energy SEPs. SEP release times with respect to event flares, as well as the event rise times, show no simple dependence on the observer's connection angle, suggesting that...

physics.space-ph↗

arXiv2017

Measurements of Forbush decreases at Mars: both by MSL on ground and by MAVEN in orbit

The Radiation Assessment Detector (RAD), on board Mars Science Laboratory's (MSL) Curiosity rover, has been measuring ground level particle fluxes along with the radiation dose rate at the surface of Mars since August 2012. Similar to neutron monitors at Earth, RAD sees many Forbush decreases (FDs) in the galactic cosmic ray (GCR) induced surface fluxes and dose rates. These FDs are associated with coronal mass ejections (CMEs) and/or stream/corotating interaction regions (SIRs/CIRs). Orbiting above the Martian atmosphere, the Mars Atmosphere and Volatile EvolutioN (MAVEN) spacecraft has also been monitoring space weather conditions at Mars since September 2014. The penetrating particle flux channels in the Solar Energetic Particle (SEP) instrument onboard MAVEN can also be employed to detect FDs. For the first time, we study the statistics and properties of a list of FDs observed in-situ at Mars, seen both on the surface by MSL/RAD and in orbit detected by the MAVEN/SEP instrument. Such a list of FDs can be used for studying interplanetary CME (ICME) propagation and SIR evolution through the inner heliosphere. The magnitudes of different FDs can be well-fitted by a power-law distribution. The systematic difference between the magnitudes of the FDs within and outside the Martian atmosphere may be mostly attributed to the energy-dependent modulation of the GCR particles by both the pass-by ICMEs/SIRs and the Martian atmosphere.

astro-ph.SR↗

arXiv2017

Catalogue of 55-80 MeV solar proton events extending through solar cycles 23 and 24

We present a new catalogue of solar energetic particle events near the Earth, covering solar cycle 23 and the majority of solar cycle 24 (1996-2016), based on the 55-80 MeV proton intensity data gathered by the SOHO/ERNE experiment. In addition to ERNE proton and heavy ion observations, data from the ACE/EPAM (near-relativistic electrons), SOHO/EPHIN (relativistic electrons), SOHO/LASCO (coronal mass ejections, CMEs), and GOES soft X-ray experiments are also considered and the associations between the particle and CME/X-ray events deduced to obtain a better understanding of each event. A total of 176 SEP events have been identified as having occurred during the time period of interest; their onset and solar release times have been estimated using both velocity dispersion analysis (VDA) and time-shifting analysis (TSA) for protons, as well as TSA for near-relativistic electrons. Additionally, a brief statistical analysis has been performed on the VDA and TSA results, as well as the X-rays and CMEs associated with the proton/electron events, both to test the viability of the VDA and to investigate possible differences between the two solar cycles. We find, in confirmation of a number of previous studies, that VDA results for protons that yield an apparent path length of 1 AU < s <~ 3 AU seem to be useful, but those outside this range are probably unreliable, as evidenced by the anticorrelation between apparent path length and release time estimated from the X-ray activity. It also appears that even the first-arriving energetic protons apparently undergo significant pitch angle scattering in the interplanetary medium, with the resulting apparent path length being on average about twice the length of the spiral magnetic field. The analysis indicates an increase in high-energy SEP events originating from the far eastern solar hemisphere; e.g., such an event...

physics.space-ph↗