Search

Face anti-spoofing (FAS) plays a vital role in preventing face recognition systems from presentation attacks. Existing face anti-spoofing datasets lack diversity due to the insufficient identity and insignificant variance, which limits the generalization ability of FAS model. In this paper, we propose Dual Spoof Disentanglement Generation (DSDG) framework to tackle this challenge by “anti-spoofing via generation”. Depending on the interpretable factorized latent disentanglement in Variational Autoencoder (VAE), DSDG learns a joint distribution of the identity representation and the spoofing pattern representation in the latent space. Then, large-scale paired live and spoofing images can be generated from random noise to boost the diversity of the training set. However, some generated face images are partially distorted due to the inherent defect of VAE. Such noisy samples are hard to predict precise depth values, thus may obstruct the widely-used depth supervised optimization. To tackle this issue, we further introduce a lightweight Depth Uncertainty Module (DUM), which alleviates the adverse effects of noisy samples by depth uncertainty learning. DUM is developed without extra-dependency, thus can be flexibly integrated with any depth supervised network for face anti-spoofing. We evaluate the effectiveness of the proposed method on five popular benchmarks and achieve state-of-the-art results under both intra- and inter- test settings. The codes are available at https://github.com/JDAI-CV/FaceX-Zoo/tree/main/addition_module/DSDG. [ABSTRACT FROM AUTHOR]

An attempt was made to drift nanoalumina particles into concrete pores through electrokinetic treatment. An external electric current (current density = 3 A/m 2 ) was applied for 3 and 15 days in reinforced concrete blocks toward steel reinforcement and microstructural characterizations (i.e. morphology observation and porosity analysis) were performed on concrete fragments of different depth from exposure surface. The morphology observation evidenced transport of nanoalumina from the exposure surface even reaching the rebar-concrete interface (up to 25–30 mm, in 15 days treatment). The porosity analysis of treated samples revealed that reduction of porosity of rebar interface was more pronounced as compared to the exposure surface and the treatment for 15 days was more beneficial for porosity refinement than treatment for 3 days. Effects of the electrokinetic NA treatment on strength of rebar-concrete interface were evaluated through pull-out test. The results showed that by increasing current density, bond strength of rebar-concrete interface increased. [ABSTRACT FROM AUTHOR]

Owing to their specific sedimentary or weathering formation history, most soils confronted in civil engineering can be categorised into clayey and sandy soils. In view of the great mechanical difference, particularly the distinct liquefaction resistance, then precisely simulate the mechanical and liquefaction characteristics of different type soils become an important research issue in seismic design of civil engineering structures. Considering soil turns into overconsolidated state once experiences stress unloading, and which affects the mechanical and liquefaction characteristics in turn. Based on the classical Cam-clay model, with the introducing of subloading surface concept, the subloading surface Cam-clay model provide an unified constitutive model in the numerical simulation of different type soils. Based on the comprehensive parametric studies of state variables evolution and anti-liquefaction capability of soil conducted in this paper, we creatively unified the relationship between OCR and relative density of soil that widely adopted in the liquefaction resistance evaluation of sandy soil, and our test results shows high correlation between these two parameters. These findings provide a solid theoretic support and guidance for the dynamic compaction method that widely adopted in anti-liquefaction design. [ABSTRACT FROM AUTHOR]