ALBERT

Description: Human behaviour analysis has introduced several challenges in various fields, such as applied information theory, affective computing, robotics, biometrics and pattern recognition [...]

Description: Thermal cameras are popular in detection for their precision in surveillance in the dark and for privacy preservation. In the era of data driven problem solving approaches, manually finding and annotating a large amount of data is inefficient in terms of cost and effort. With the introduction of transfer learning, rather than having large datasets, a dataset covering all characteristics and aspects of the target place is more important. In this work, we studied a large thermal dataset recorded for 20 weeks and identified nine phenomena in it. Moreover, we investigated the impact of each phenomenon for model adaptation in transfer learning. Each phenomenon was investigated separately and in combination. the performance was analyzed by computing the F1 score, precision, recall, true negative rate, and false negative rate. Furthermore, to underline our investigation, the trained model with our dataset was further tested on publicly available datasets, and encouraging results were obtained. Finally, our dataset was also made publicly available.

Description: Threat assessments continue to conclude that terrorist groups and individuals as well as those wanting to cause harm to society have the ambition and increasing means to acquire unconventional weapons such as improvised nuclear explosive devices and radiological disposal devices. Such assessments are given credence by public statements of intent by such groups/persons, by reports of attempts to acquire radioactive material and by law enforcement actions which have interdicted, apprehended or prevented attempts to acquire such material. As a mechanism through which to identify radioactive materials being transported on an individual’s person, this work sought to develop a detection system that is of lower-cost, reduced form-factor and more covert than existing infrastructure, while maintaining adequate sensitivity and being retrofittable into an industry standard and widely utilised Gunnebo Speed Gate system. The system developed comprised an array of six off-set Geiger–Muller detectors positioned around the gate, alongside a single scintillator detector for spectroscopy, triggered by the systems inbuilt existing IR proximity sensor. This configuration served to not only reduce the cost for such a system but also allowed for source localisation and identification to be performed. Utilising the current setup, it was possible to detect a 1 µSv/h source carried into the Speed Gate in all test scenarios, alongside locating and spectrally analysing the material in a significant number.

Description: Chitin is one of the most abundant biomolecules on earth, occurring in crustacean shells and cell walls of fungi. While the polysaccharide is threatening to pollute coastal ecosystems in the form of accumulating shell-waste, it has the potential to be converted into highly profitable derivatives with applications in medicine, biotechnology, and wastewater treatment, among others. Traditionally this is still mostly done by the employment of aggressive chemicals, yielding low quality while producing toxic by-products. In the last decades, the enzymatic conversion of chitin has been on the rise, albeit still not on the same level of cost-effectiveness compared to the traditional methods due to its multi-step character. Another severe drawback of the biotechnological approach is the highly ordered structure of chitin, which renders it nigh impossible for most glycosidic hydrolases to act upon. So far, only the Auxiliary Activity 10 family (AA10), including lytic polysaccharide monooxygenases (LPMOs), is known to hydrolyse native recalcitrant chitin, which spares the expensive first step of chemical or mechanical pre-treatment to enlarge the substrate surface. The main advantages of enzymatic conversion of chitin over conventional chemical methods are the biocompability and, more strikingly, the higher product specificity, product quality, and yield of the process. Products with a higher Mw due to no unspecific depolymerisation besides an exactly defined degree and pattern of acetylation can be yielded. This provides a new toolset of thousands of new chitin and chitosan derivatives, as the physio-chemical properties can be modified according to the desired application. This review aims to provide an overview of the biotechnological tools currently at hand, as well as challenges and crucial steps to achieve the long-term goal of enzymatic conversion of native chitin into specialty chemical products.

Description: The effect on the gas permeance properties and structural morphology of the presence of methyl functional groups in a silica membrane was studied. Membranes were synthesized via chemical vapor deposition (CVD) at 650 °C and atmospheric pressure using three silicon compounds with differing numbers of methyl- and methoxy-functional groups: tetramethyl orthosilicate (TMOS), methyltrimethoxysilane (MTMOS), and dimethyldimethoxysilane (DMDMOS). The residence time of the silica precursors in the CVD process was adjusted for each precursor and optimized in terms of gas permeance and ideal gas selectivity criteria. Final H2 permeances at 600 °C for the TMOS-, MTMOS-, and DMDMOS-derived membranes were respectively 1.7 × 10−7, 2.4 × 10−7, and 4.4 × 10−8 mol∙m−2∙s−1∙Pa−1 and H2/N2 selectivities were 990, 740, and 410. The presence of methyl groups in the membranes fabricated with the MTMOS and DMDMOS precursors was confirmed via Fourier-transform infrared (FTIR) spectroscopy. From FTIR analysis, an increasing methyl signal in the silica structure was correlated with both an improvement in the hydrothermal stability and an increase in the apparent activation energy for hydrogen permeation. In addition, the permeation mechanism for several gas species (He, H2, Ne, CO2, N2, and CH4) was determined by fitting the gas permeance temperature dependence to one of three models: solid state, gas-translational, or surface diffusion.

Description: With the consolidation of the new data protection regulation paradigm for each individual within the European Union (EU), major biometric technologies are now confronted with many concerns related to user privacy in biometric deployments. When individual biometrics are disclosed, the sensitive information about his/her personal data such as financial or health are at high risk of being misused or compromised. This issue can be escalated considerably over scenarios of non-cooperative users, such as elderly people residing in care homes, with their inability to interact conveniently and securely with the biometric system. The primary goal of this study is to design a novel database to investigate the problem of automatic people recognition under privacy constraints. To do so, the collected data-set contains the subject’s hand and foot traits and excludes the face biometrics of individuals in order to protect their privacy. We carried out extensive simulations using different baseline methods, including deep learning. Simulation results show that, with the spatial features extracted from the subject sequence in both individual hand or foot videos, state-of-the-art deep models provide promising recognition performance.

Description: Efficient and robust evaluation of kernel processing from corn silage is an important indicator to a farmer to determine the quality of their harvested crop. Current methods are cumbersome to conduct and take between hours to days. We present the adoption of two deep learning-based methods for kernel processing prediction without the cumbersome step of separating kernels and stover before capturing images. The methods show that kernels can be detected both with bounding boxes and at pixel-level instance segmentation. Networks were trained on up to 1393 images containing just over 6907 manually annotated kernel instances. Both methods showed promising results despite the challenging setting, with an average precision at an intersection-over-union of 0.5 of 34.0% and 36.1% on the test set consisting of images from three different harvest seasons for the bounding-box and instance segmentation networks respectively. Additionally, analysis of the correlation between the Kernel Processing Score (KPS) of annotations against the KPS of model predictions showed a strong correlation, with the best performing at r(15) = 0.88, p = 0.00003. The adoption of deep learning-based object recognition approaches for kernel processing measurement has the potential to lower the quality assessment process to minutes, greatly aiding a farmer in the strenuous harvesting season.

Description: In response to the Fukushima Daiichi Nuclear Power Plant accident, there has occurred the unabated growth in the number of airborne platforms developed to perform radiation mapping—each utilising various designs of a low-altitude uncrewed aerial vehicle. Alongside the associated advancements in the airborne system transporting the radiation detection payload, from the earliest radiological analyses performed using gas-filled Geiger-Muller tube detectors, modern radiation detection and mapping platforms are now based near-exclusively on solid-state scintillator detectors. With numerous varieties of such light-emitting crystalline materials now in existence, this combined desk and computational modelling study sought to evaluate the best-available detector material compatible with the requirements for low-altitude autonomous radiation detection, localisation and subsequent high spatial-resolution mapping of both naturally occurring and anthropogenically-derived radionuclides. The ideal geometry of such detector materials is also evaluated. While NaI and CsI (both elementally doped) are (and will likely remain) the mainstays of radiation detection, LaBr3 scintillation detectors were determined to possess not only a greater sensitivity to incident gamma-ray radiation, but also a far superior spectral (energy) resolution over existing and other potentially deployable detector materials. Combined with their current competitive cost, an array of three such composition cylindrical detectors were determined to provide the best means of detecting and discriminating the various incident gamma-rays.

Description: The challenge of getting machines to understand and interact with natural objects is encountered in important areas such as medicine, agriculture, and, in our case, slaughterhouse automation. Recent breakthroughs have enabled the application of Deep Neural Networks (DNN) directly to point clouds, an efficient and natural representation of 3D objects. The potential of these methods has mostly been demonstrated for classification and segmentation tasks involving rigid man-made objects. We present a method, based on the successful PointNet architecture, for learning to regress correct tool placement from human demonstrations, using virtual reality. Our method is applied to a challenging slaughterhouse cutting task, which requires an understanding of the local geometry including the shape, size, and orientation. We propose an intermediate five-Degree of Freedom (DoF) cutting plane representation, a point and a normal vector, which eases the demonstration and learning process. A live experiment is conducted in order to unveil issues and begin to understand the required accuracy. Eleven cuts are rated by an expert, with 8 / 11 being rated as acceptable. The error on the test set is subsequently reduced through the addition of more training data and improvements to the DNN. The result is a reduction in the average translation from 1.5 cm to 0.8 cm and the orientation error from 4 . 59 to 4 . 48 . The method’s generalization capacity is assessed on a similar task from the slaughterhouse and on the very different public LINEMOD dataset for object pose estimation across view points. In both cases, the method shows promising results. Code, datasets, and supplementary materials are available at https://github.com/markpp/PoseFromPointClouds.

Description: In the condensed liquid phase, both single- and multicomponent Lennard–Jones (LJ) systems obey the “hidden-scale-invariance” symmetry to a good approximation. Defining an isomorph as a line of constant excess entropy in the thermodynamic phase diagram, the consequent approximate isomorph invariance of structure and dynamics in appropriate units is well documented. However, although all measures of the structure are predicted to be isomorph invariant, with few exceptions only the radial distribution function (RDF) has been investigated. This paper studies the variation along isomorphs of the nearest-neighbor geometry quantified by the occurrence of Voronoi structures, Frank–Kasper bonds, icosahedral local order, and bond-orientational order. Data are presented for the standard LJ system and for three binary LJ mixtures (Kob–Andersen, Wahnström, NiY2). We find that, while the nearest-neighbor geometry generally varies significantly throughout the phase diagram, good invariance is observed along the isomorphs. We conclude that higher-order structural correlations are no less isomorph invariant than is the RDF.