In this paper, we report on shear flows in domains that contain a macroscopic interface between a highly porous medium and a pure fluid. Our study is based on the single-domain approach, according to which, the same set of governing equations is employed for both inside the porous medium and in the pure-fluid domain. In particular, we introduce a mathematical model for the flows of interest that is derived directly from a continuum theory for fluid-saturated granular materials. The resulting set of equations is a variation of the well-known unsteady Darcy-Brinkman model. First, we employ this model to perform a linear stability analysis of inviscid shear layers over a highly porous medium. Our analysis shows that such layers are unconditionally unstable. Next, we present results from numerical simulations of temporally evolving shear layers in both two and three dimensions. The simulations are performed via a recently designed algorithm that employs a predictor-corrector time-marching scheme and a projection method for the computation of the pressure field on a collocated grid. According to our numerical predictions, the onset of the Kelvin-Helmholtz instability leads to the formation of vortices that extend to both sides of the material interface, thus producing substantial recirculation inside the porous medium. These vortices eventually merge, leading to significant growth of the shear layer and, in three dimensional flows, transition to turbulence. The dynamics of the shear layers, including growth rate and self-similarity, is presented and analysed. Finally, the structure of these layers is described in detail and compared to the one of plain mixing layers.

Over the past few years, deep learning methods have shown remarkable results in many face-related tasks including automatic facial expression recognition (FER) in-thewild. Meanwhile, numerous models describing the human emotional states have been proposed by the psychology community. However, we have no clear evidence as to which representation is more appropriate and the majority of FER systems use either the categorical or the dimensional model of affect. Inspired by recent work in multi-label classification, this paper proposes a novel multi-task learning (MTL) framework that exploits the dependencies between these two models using a Graph Convolutional Network (GCN) to recognize facial expressions in-the-wild. Specifically, a shared feature representation is learned for both discrete and continuous recognition in a MTL setting. Moreover, the facial expression classifiers and the valence-arousal regressors are learned through a GCN that explicitly captures the dependencies between them. To evaluate the performance of our method under real-world conditions we train our models on AffectNet dataset. The results of our experiments show that our method outperforms the current state-of-the-art methods on discrete FER.

No abstract

No abstract

No abstract