Coin grading is a crucial process in numismatics that determines a coin’s value on the basis of its condition. However, the traditional grading procedure is performed by human which makes it subjective and time-consuming, with different experts potentially assigning different grades to the same coin. This paper introduces a Deep Learning (DL) model for the type-invariant automatic coin grading across the full 70-points Sheldon scale. We propose a Siamese neural network based on EfficientNet encoders for dual-sided coin evaluation, which performs the analysis of both the obverse and reverse images of coins. The model is trained and evaluated on a comprehensive dataset containing 36,205 unique professionally graded samples covering all 30 grades from Poor (P)1 to Mint State (MS)70 across the multiple coin types. The proposed DL solution achieves 0.3403 accuracy and 1.52 Mean Absolute Error (MAE) along the Sheldon scale, outperforming both the Machine Learning (ML) baseline and the state-of-the-art approaches. Unlike earlier studies that were limited to fewer coin types or reduced grading scales, our work addresses the full Sheldon scale grading across the multiple coin types, providing a more robust and generalizable solution for automatic coin grading.

Computer vision systems have been integrated into facilities dealing with the sorting of household waste. This solution allows for the sorting efficiency improvement and cost reduction. However, challenges associated with the poor annotation quality of existing waste segmentation datasets, unsuitable environment for recognition on a conveyor belt, or limited data for creating an effective and cost-efficient sorting system using visible range cameras significantly limit the application efficiency of computer vision systems. In this article, we report on the data-centric pipeline for enhancing the precision of predictions in multiclass household waste segmentation on a conveyor belt. In particular, we have demonstrated that by employing a pseudo-annotation approach combined with an object-based data augmentation algorithm, it is possible to train a model on a set of 'simple' images and achieve satisfactory results when estimating the model on a set of 'complex' images. We collected and prepared the dataset consisting of 5 k manually labeled data and additionally 10 k pseudo-labeled data by object-based augmentation. The proposed pipeline incorporates data balancing, transfer learning, and pseudo-labeling to improve the mean Average Precision (mAP) of the YOLOV8 segmentation model from 67 % to 83 % for 'simple' use case scenarios and from 42 % to 59 % or 'complex' industrial solutions.

We report the results of systematic ab initio modelling of various configurations of iron and cobalt impurities embedded in the (110), (101), and (100) surfaces of anatase TiO2, with and without oxygen vacancies. The simulation results demonstrate that incorporation into interstitial voids at the surface level is significantly more favourable than other configurations for both iron and cobalt. The calculations also demonstrate the crucial effect of the facet as well as the lesser effects of other factors, such as vacancies and strain on the energetics of defect incorporation, magnetic moment, bandgap, and catalytic performance. It is further shown that there is no tendency towards the segregation or clustering of impurities on the surface. The calculated free energies of the hydrogen evolution reaction in acidic media predict that iron impurities embedded in the (101) surface of anatase TiO2 can be a competitive catalyst for this reaction.

Crying is an inevitable character trait that occurs throughout the growth of infants, under conditions where the caregiver may have difficulty interpreting the underlying cause of the cry. Crying can be treated as an audio signal that carries a message about the infant's state, such as discomfort, hunger, and sickness. The primary infant caregiver requires traditional ways of understanding these feelings. Failing to understand them correctly can cause severe problems. Several methods attempt to solve this problem; however, proper audio feature representation and classifiers are necessary for better results. This study uses time-, frequency-, and time-frequency-domain feature representations to gain in-depth information from the data. The time-domain features include zero-crossing rate (ZCR) and root mean square (RMS), the frequency-domain feature includes the Mel-spectrogram, and the time-frequency-domain feature includes Mel-frequency cepstral coefficients (MFCCs). Moreover, time-series imaging algorithms are applied to transform 20 MFCC features into images using different algorithms: Gramian angular difference fields, Gramian angular summation fields, Markov transition fields, recurrence plots, and RGB GAF. Then, these features are provided to different machine learning classifiers, such as decision tree, random forest, K nearest neighbors, and bagging. The use of MFCCs, ZCR, and RMS as features achieved high performance, outperforming state of the art (SOTA). Optimal parameters are found via the grid search method using 10-fold cross-validation. Our MFCC-based random forest (RF) classifier approach achieved an accuracy of 96.39%, outperforming SOTA, the scalogram-based shuffleNet classifier, which had an accuracy of 95.17%.

Parkinson's disease (PD) causes physical activity loss, tremors, stiffness, and coordination-related issues. The state-of-the-art studies utilize complex experimental testbeds for data acquisition and consequent application of Machine learning (ML) methods for the PD diagnosis. However, their performance is still the subject to improve. This research aims to distinguish PD from healthy control (HC) using four wearable sensors and ML, where two sensors are placed on each hand, wrist, and dorsum. The dataset of 54 patients, 21 with HC and 33 with PD, was collected in a hospital while the patient performed 11 exercises under the supervision of neurologists. The dataset was preprocessed and segmented into overlapping windows, and ML algorithms were used in terms of analysis of frequency ranges, features, and exercises. This study demonstrates a simple and accurate method for detecting PD in clinical or home settings with an average of 0.936 $f 1_{{micro }}$ and 0.935 ROC while the performance of ‘best’ exercises achieves 100% ${f}1_{micro}$ .

In this work, we address the cheating problem in video games and provide instruments for reliable detection of various cheaters based on their in-game behavior in the most popular FPS game Counter-Strike: Global Offensive. For this purpose, we collect more than 2,243 competitive game records through official Valve servers. This corresponds to more than 14,000 players with various gaming skill levels. Thorough application of mathematical processing methods, as well as the eSports expert assistance, allows us to develop meaningful and accurate metrics for measuring in-game player actions. We report that with their help cheater detection has high accuracy of 85%, and outperform VAC, VACNet, and OverWatch in a comparative analysis. The tools are additionally tested on the professional players and show a low false positive rate.

An incredibly wide range of plastic products and enormous volumes of generated plastic wastes makes sorting technologies highly important. In order to enhance the efficiency and accuracy of sorting process, this article proposes an Internet of Video Things solution based on the deep learning algorithms for image recognition of plastic waste on a moving conveyor belt and embedded intelligence. In particular, the state-of-the-art object detection models, including Faster R-CNN, RetinaNet and YOLOv8 are used. Target categories of plastic are Polyethylene Terephthalate (PET) and Polypropylene (PP). Furthermore, we implemented quantization techniques for trained models on a commercial off-the-shelf embedded system for fast processing time. We achieved a high mean Average Precision (mAP) metric of 77.74% and accuracy of 95.67% on a test set and are fine-tuned and optimized for the deployment on an Nano embedded system providing 20 frames per second. This research contributes to the field of application of Internet of Things by demonstrating the efficacy of deep learning algorithms run on the embedded system in the industrial plastic waste sorting process. The findings highlight the practical applicability of these algorithms and offer insights into resource management and recycling practices.

This International Women in Engineering Day (INWED 2024) Communications Engineering celebrates women in innovative startups.

2D photodetectors can considerably outperform their Si counterparts and thus appear promising for contemporary optoelectronic circuits. However, limited charge separation efficiency and poor optical absorption make it complicated to create ultrasensitive broad response photodetectors using a single 2D material. While hybrid heterostructures combining 2D and 1D materials could be a promising solution, previously used binary/binary 1D/2D photodetectors have a performance that is merely satisfactory at best. To address these limitations, here we for the first time demonstrate vertically stacked ternary/binary photodetectors (VPDs) based on 1D ternary CdSxSe1-x nanoribbon (NR)/2D binary PbI2 nanosheets (NS) heterostructures. While excellent crystallinity of the ternary/binary structure enables efficient carrier transport, a type-II heterojunction at the interfaces facilitates charge separation. As a result, our devices exhibit excellent photodetection ability in broad range with superior photoresponsivity (567 A/W), photosensitivity (1.60 x 10(7) %), specific detectivity (1.96 x 10(15) Jones), external quantum efficiency (1.17 x 10(5) %), and the response speed which is several orders of magnitude higher as compared to our reference devices based on PbI2 NS and most of the previously reported photodetectors in visible range. Furthermore, our VPDs fabricated on flexible substrates show stable operation after being subjected to multiple bending cycles and retain excellent performance following several weeks of storage in an air environment. Our results provide advanced insights into the design and fabrication of environmentally stable flexible VPDs for future optoelectronics circuits.

A number of polyolefin plastics including Polypropylene (PP), Polyethylene (PE), Polystyrene (PS) are widespread plastics worldwide. At the same time, they are frequently recycled plastics. Still a considerable amount of waste plastics come in a mixed mode imposing technological challenges for recycling. Households cannot provide thorough separation on the early stage, especially if the plastic products are not marked properly. In this paper, we report on the analysis of two pair of plastic shavers of the similar color and shape while made of different plastics. For this reason an experimental testbed based on a Raman Spectrometer with an embedded microscope has been designed. The proposed solution enables the identification of PP and PS by the sensor system with the singular spectral range 1320 nm - 1740 nm. There is an opportunity to measure three spectral ranges simultaneously.

Our review seeks to elucidate the current state-of-the-art in studies of 70-kilodalton-weighed heat shock proteins (Hsp70) in neurodegenerative diseases (NDs). The family has already been shown to play a crucial role in pathological aggregation for a wide spectrum of brain pathologies. However, a slender boundary between a big body of fundamental data and its implementation has only recently been crossed. Currently, we are witnessing an anticipated advancement in the domain with dozens of studies published every month. In this review, we briefly summarize scattered results regarding the role of Hsp70 in the most common NDs including Alzheimer’s disease (AD), Parkinson’s disease (PD), and amyotrophic lateral sclerosis (ALS). We also bridge translational studies and clinical trials to portray the output for medical practice. Available options to regulate Hsp70 activity in NDs are outlined, too.

In this paper, we report on a dominant hand invariant 1-D convolution model for distinguishing between Parkinson’s disease (PD) and healthy subjects. For realizing this approach, we propose an STFT-based method for IMU data axes fusion. We learn how both the different frequency ranges and the period of the STFT window affect the efficiency of Parkinson’s disease detection. We test this solution on the dataset collected from 58 subjects. Our results show superiority of the proposed axes fusion method in 70% of cases in comparison with the state-of-the-art. Results also prove the efficiency of the proposed 1-D convolution hand invariant model with the best scores 98% of AUC and 92% of F1 and accuracy metrics. In addition, we show the STFT window must be at least 2 seconds of length, while the frequency range must include the frequencies below 3 Hz.

E-sports is the state-of-the-art digital sport where players and teams compete against each other in various gaming disciplines. The most significant part of the industry includes tournaments specifically in multiplayer team-based games where outstanding individual performance is not enough and where a team is perceived as a single mechanism. In such cases, the game performance should not be limited to the individual team member contributions and take into account internal players’ interactions. It is essential to emphasize that e-sports players are humans, therefore, while evaluating their performance it is necessary to take into account the psychological and emotional points. In this work, we propose an approach to the game performance analysis focusing on the factors insufficiently investigated earlier, such as the team interaction efficiency in terms of auxiliary in-game attack tool usage, as well as the teammates’ emotional state described by an arousal-valence model. First, we perform the analysis of a large amount of in-game Counter Strike:Global Offensive discipline tournament data, derive detailed metrics related to the team actions and show their strong correlation with the game outcome characterized by a high level of significance (p-value <0.001). Second, we train the regression models separately for the valence and arousal scales with RMSE 0.497 and 0.561, respectively, and prove their consistency by validation on E-sports audio data. Our findings show that the team performance significantly depends both on the effectiveness of the coordinated grenade use and on the players’ emotional state during the game.

We propose and demonstrate a novel range of models to accurately determine the optical properties of nitrogen-free carbon quantum dots (CQDs) with ordered graphene layered structures. We confirm the results of our models against the full range of experimental results for CQDs available from an extensive review of the literature. The models can be equally applied to CQDs with varied sizes and with different oxygen contents in the basal planes of the constituent graphenic sheets. We demonstrate that the experimentally observed blue fluorescent emission of nitrogen-free CQDs can be associated with either small oxidised areas on the periphery of the graphenic sheets, or with sub-nanometre non-functionalised islands of sp(2)-hybridised carbon with high symmetry confined in the centres of oxidised graphene sheets. Larger and/or less symmetric non-functionalised regions in the centre of functionalised graphene sheet are found to be sources of green and even red fluorescent emission from nitrogen-free CQDs. We also demonstrate an approach to simplify the modelling of the discussed sp(2)-islands by substitution with equivalent strained polycyclic aromatic hydrocarbons. Additionally, we show that the bandgaps (and photoluminescence) of CQDs are not dependent on either out-of-plane corrugation of the graphene sheet or the spacing between sp(2)-islands. Advantageously, our proposed models show that there is no need to involve light-emitting polycyclic aromatic molecules (nanographenes) with arbitrary structures grafted to the particle periphery to explain the plethora of optical phenomena observed for CQDs across the full range of experimental works.

Internet of Things (IoT) solutions have greatly evolved recently with the application of Artificial Intelligence (AI). Indeed, AI enriches the IoT with intelligence capabilities. In particular, AI methods are highly effective in the scope of the Internet of Medical Things (IoMT) for the applications requiring decision support for doctors. In this work, we propose a Deep Learning (DL) framework for the classification of Parkinson's disease (PD) and Progressive Supranuclear Palsy (PSP). In contrast to the state-of-the-art solutions relying on only the saccade test for the classification of these neurodegenerative diseases, we collect a dataset while the subjects perform five exercises including saccade, spontaneous nystagmus, optokinetic nystagmus, pursuit, and gaze test. We then extract the pupil features (coordinates, area, minor axis, and major axis) using the image segmentation DL model and represent them as images using Gramian Angular Difference Field (GADF) time series imagining algorithm. The resultant images are supplied to the proposed disease-detection model for running the classification procedure. The best classification results were obtained for the optokinetic exercise with the accuracy of 96.9%, 90.8%, and 96.9% for the left, right, and both eyes, respectively.

At present, there is a significant focus on improving agricultural productivity, which is essential both on a national and global scale. There are many factors that impact the quality and quantity of crop yields, and environmental changes such as weather conditions require careful consideration and quick action to avoid significant losses. When it comes to managing fruit orchards, large areas need to be monitored, and unmanned aerial vehicles (UAVs) are currently being used to automate and enhance the process. UAVs can capture images that can be analyzed using neural-based methods to extract useful information about the state of the trees in the orchard. However, there are some challenges that must be addressed, such as the scarcity of open-access labeled datasets and the imbalanced distribution of target classes, which include rare events or anomalous vegetation states. To tackle these issues, in this paper, a unique dataset of apple trees observations captured by UAVs has been collected and shared to the research community. This dataset includes healthy and unhealthy trees with formed and unformed crowns. Experiments were conducted using YOLOv5 neural network for object detection to evaluate its effectiveness in solving agricultural remote sensing tasks. To adjust model’s performance, we proposed a task-specific transfer learning approach that involves pre-training the model on a synthetic dataset. The synthetic dataset was generated using object-based augmentation (OBA) with the original target objects. Thereby, instead of utilizing pre-trained weights from the general domain such as COCO dataset, we use not only domain-specific UAV-derived data for pre-training but the images for the same task of apple tree health examining. The proposed approach allowed us to increase the mAP from 0.642 to 0.706 compared with the conventional approach.

École Nationale Supérieure de Chimie de Paris, ParisTech, France 7-10 December 2021 Smart nanomaterials are the basis of diverse emerging applications, including wearable and printed electronics, flexible optoelectronics, CMOS photonics, quantum computing, smart coatings and thin films. This collection focuses on the most recent scientific and technological advances, innovations and new practical applications of novel smart nanomaterials presented at the 4th Smart Nanomaterials Conference, SNAIA 2021. It covers topics ranging from advances in the most critical aspects in chemistry and material fabrication of nanomaterials, to the engineering of prototype devices and systems. List of Scientific Committee, Organising Committee, Editorial Board, Plenary Speakers, Keynote Speakers and Invited Speakers are available in this pdf.

Unusual optical anisotropy was experimentally observed in hexagonal boron nitride thin films produced from bulk boron nitride via ultrasonication. Both the linear and circular polarisation demonstrated a well-defined single axis of anisotropy over a large sample area. To understand this phenomenon, we employed statistical analysis of optical microscopy images and atomic force microscopy to reveal an ordered particle density distribution at the microscopic level corresponding to the optical axis observed in the polarisation data. The direction of the observed ordering matched the axis of anisotropy. Hence, we attribute the measured optical anisotropy of the thin films to microscopic variations in the particle density distribution.

Тонкие пленки гексагонального нитрида бора толщиной несколько моноатомных слоев были изготовлены путем расщепления объемных образцов в ультразвуковой ванне. Исследовались спектры пропускания, отражения и фотолюминесценции таких пленок. Были измерены спектральные зависимости линейной и круговой поляризации прошедшего через образец света. Исследование с помощью сканирующей электронной микроскопии показало однородность полученных образцов. Однако исследование параметров Стокса прошедшего через образец света позволило установить наличие скрытой анизотропии оптических свойств этих пленок. Ключевые слова: нитрид бора, оптическая анизотропия, спектроскопия, поляризация, тонкие пленки.

Thin hexagonal boron nitride films with a thickness of several monoatomic layers have been manufactured by splitting of bulk samples in an ultrasonic bath. The transmission, reflection, and photoluminescence spectra of such films are studied. The spectral dependences of linear and circular polarization of light passing through the sample are measured. Scanning electron microscopy investigation demonstrates homogeneity of the obtained samples. However, investigation of Stokes parameters of light passing through the sample makes it possible to reveal hidden anisotropy of optical properties of these films.