Recognition System Research Articles

HDN-Net: A Hybrid Deep Neural Network to Improve Iris Recognition in Unconstrained Environments with Eyeglasses

Recently, the use of iris recognition technology for biometric authentication has gained widespread acceptance due to the rich texture of the iris region, which provides a reliable standard for recognising individuals, as well as the non-intrusive nature of this method. However, the presence of eyeglasses poses a significant challenge to the accuracy of such systems. In unrestricted environments, current iris recognition techniques cannot effectively extract distinguishing features of the iris. Eyeglasses introduce scratches, specular reflections, dirt, blurriness, and other noise factors over the image of the iris, resulting in low recognition accuracy. To tackle this challenge, researchers have proposed the HDN-Net architecture. This architecture employs a multi-CNN model to combine the features of both the right and left iris images, extracting more distinguishing features to improve the accuracy of the classification task in the presence of challenges caused by eyeglasses. Experiment results show that the proposed iris recognition system achieves more promising performance compared to previous methods used in this field. The overall performance of our suggested HDN-Net method on the UBIRIS.V2 and CASIA-Iris.V4-1000 databases achieved 97.89% and 98.79% accuracy, respectively. Thus, the proposed HDN-Net method consistently outperforms other traditional and deep learning approaches and has the possibility to improve the accuracy of iris recognition systems (IRS) in real-world scenarios.

Emotion Recognition Using EEG Signals and Audiovisual Features with Contrastive Learning.

Multimodal emotion recognition has emerged as a promising approach to capture the complex nature of human emotions by integrating information from various sources such as physiological signals, visual behavioral cues, and audio-visual content. However, current methods often struggle with effectively processing redundant or conflicting information across modalities and may overlook implicit inter-modal correlations. To address these challenges, this paper presents a novel multimodal emotion recognition framework which integrates audio-visual features with viewers' EEG data to enhance emotion classification accuracy. The proposed approach employs modality-specific encoders to extract spatiotemporal features, which are then aligned through contrastive learning to capture inter-modal relationships. Additionally, cross-modal attention mechanisms are incorporated for effective feature fusion across modalities. The framework, comprising pre-training, fine-tuning, and testing phases, is evaluated on multiple datasets of emotional responses. The experimental results demonstrate that the proposed multimodal approach, which combines audio-visual features with EEG data, is highly effective in recognizing emotions, highlighting its potential for advancing emotion recognition systems.

An IoT solution that detects human physical activities to support health assessment based on sensors and supervised learning algorithms

At the moment, human physical activities can be detected by activity recognition systems to help with health status evaluation. One of the key issues in human activity recognition (HAR) systems is cost, speed, and accuracy. With the advancement of mobile devices and increasingly sophisticated microprocessors, along with the integration of various sensors including accelerometers, human activity recognition has become easier and faster. However, the accuracy of activity classification based on accelerometer data depends on many different factors, which can lead to potential deviations in results. Through various experiments, we have found that selecting optimal features is crucial for the performance of classification. In this paper, we propose a simple feature set that yields very promising results with an accuracy of up to 99%.

A NIRS-based recognition of coal and rock using convolution-multiview broad learning system

Achieving high production in the top coal caving process from thick coal seams is crucial. Thus, the timely decision of when to stop caving poses an urgent challenge to impact the mining loss rate and cost recovery. To address this issue, an innovative recognition system has been developed using Near-Infrared Spectroscopy (NIRS) technology. It stands out for its on-site usability, it enables rapid data collection and local recognition at the longwall face. Furthermore, to overcome the limitations of existing methods in adapting to variations in spectral data quality during on-site collection and the lack of integration of spectral data across different feature processing stages, a coal-rock recognition method has been developed which can ignore the influence of acquisition factors(granularity, light source angle, and detection sensor angle). This method incorporates the features of convolution and multi-view into the BLS model, the designed model structure exhibits a remarkable recognition accuracy of 99.78 %. The model was deployed into the recognition system, and experimental tests were conducted on the working face. The results showed that the recognition system can effectively identify the entire coal-caving process and achieve a recognition accuracy of 92.3 %. This capability is crucial for determining the optimal point to stop roof caving.

Soft set-based MSER end-to-end system for occluded scene text detection, recognition and prediction

The presence of unpredictable occlusions on natural scene text is a significant challenge, exacerbating the difficulties already posed on text detection and recognition by the variability of such images. Addressing the need for a robust, consistently performing approach that can effectively address the above challenges, this paper presents a new Soft Set-based end-to-end system for text detection, recognition and prediction in occluded natural scene images. This is the first approach to integrate text detection, recognition and prediction, unlike existing systems developed for end-to-end text spotting (text detection and recognition) only. For candidate text components detection, the proposed combination of Soft Sets with Maximally Stable Extremal Regions (SS-MSER) improves text detection and spotting in natural scene images, irrespectively of the presence of arbitrarily orientated and shaped text, complex backgrounds and occlusion. Furthermore, a Graph Recurrent Neural Network is proposed for grouping candidate text components into text lines and for fitting accurate bounding boxes to each word. Finally, a Convolutional Recurrent Neural Network (CRNN) is proposed for the recognition of text and for predicting missing characters due to occlusion. Experimental results on a new occluded scene text dataset (OSTD) and on the most relevant benchmark natural scene text datasets demonstrate that the proposed system outperforms the state-of-the-art in text detection, recognition and prediction. The code and dataset are available at https://github.com/alloydas/Softset-MSER-Based-Occluded-Scene-Text-Spotting/blob/master/Soft_set_MSER.ipynb

Effect of Recognition Practices on Employee Engagement in Fairtrade Horticultural Firms in Kenya

While most people understand that engaged employees are often happier and more satisfied in their job, the main reason that engagement is now a focal point is because it leads to real business results. Emotional states are hard to understand and measure. Employees are becoming key business drivers for any organization. The objective of the study was to determine the effect of compensation system on employee engagement in Fairtrade horticultural firms in Kenya. The study was guided by expectancy theory and social exchange theory. A cross-sectional descriptive survey research design was adopted. The study targeted 17 fairtrade horticultural firms as unit of analysis and 5,600 employees as the unit of observation. The sample size was 373 respondents drawn from low level, middle level, and top-level management. The study relied mainly on primary data. The study adopted stratified and simple random sampling techniques. After the stratification of the respondents into strata, simple random sampling was adopted to pick up the respondents. The researcher used questionnaire as the research instrument. Descriptive and inferential analysis were computed using SPSS version 26. In order to analyze the relationship between the independent variables and the dependent variable the study used regression analysis at 5% level of significance. Specifically, simple linear regression analysis. The findings indicated that recognition system has a positive and significant effect on employee engagement in Fairtrade horticultural firms in Kenya. The study concluded that recognition system contributes significantly to the employee engagement in Fairtrade horticultural firms in Kenya. From the results, the study recommends that top management horticultural firm should come up with well-defined structure that will regularly recognize employees with exemplary performance in various categories. This kind of recognition by the management as well as their supervisors would make employees feel appreciated for their input and therefore, help the organization to achieve their goals.

Proposed method for against adversarial images based on ResNet architecture

Our world is becoming increasingly automated due to the application of deep learning/machine learning models to systems, but these systems are vulnerable to adversarial attacks, which create deceptive data to trick them. Without proper defenses, attackers can exploit deep learning systems in facial recognition, self-driving cars, and social media filters. Research on adversarial image generation and methods to against attacks is important. This paper proposes employing the ResNet architecture with adversarial training to against adversarial images. The model is tested on a Hybrid CIFAR-10 dataset, which is designed to improve robustness and accuracy by incorporating GAN-generated images. The proposed model achieves an accuracy of over 95%, which is better than three other state-of-the-art architectures VGG19_bn, ShuffleNetV2, and RepVGG_a2.

Fusing YOLOv5s-MediaPipe-HRV to Classify Engagement in E-learning: from the Perspective of External Observations and Internal Factors

The rapid advancements in computer vision technology present significant potential for the automatic recognition of learner engagement in E-learning. We conducted a two-stage experiment to assess learner engagement based on behavioural (external observations) and physiological (internal factors) cues. Using computer vision technology and wearable sensors, we extracted three feature sets: action, head posture and heart rate variability (HRV). Subsequently, we integrated our constructed YOLOv5s–MediaPipe behaviour detection model with a physiological detection model based on HRV to comprehensively evaluate learners’ behavioural, affective and cognitive engagement. Additionally, we developed a method and criteria for assessing distraction based on behaviour, ultimately creating a comprehensive, efficient, low-cost and easy-to-use system for the automatic recognition of learner engagement. Experimental results showed that our improved YOLOv5s model achieved a mean average precision of 92.2%, while halving both the number of parameters and model size. Unlike other deep learning-based methods, using MediaPipe–OpenCV for head posture analysis offers advantages in real-time performance, making it lightweight and easy to deploy. Our proposed long short-term memory classifier, based on sensitive HRV metrics and their normalisation, demonstrated satisfactory performance on the test set, with an accuracy = 80%, precision = 81%, recall = 80% and an F1 score = 80%.

Enhancing the Recognition of Handwritten Arabic Characters through Hybrid Convolutional and Bidirectional Recurrent Neural Network Models

Handwritten Arabic character recognition remains a challenging task in pattern recognition due to the inherent complexities of the cursive script and visual similarities between characters. While deep learning techniques have demonstrated promising results in this domain, further enhancements to the model architecture can drive even greater performance improvements. This study introduces a hybrid deep learning approach that combines Convolutional Neural Networks (CNNs) with Bidirectional Recurrent Neural Networks, specifically Bidirectional Long Short-Term Memory (Bi-LSTM) and Bidirectional Gated Recurrent Units (Bi-GRU). By leveraging the strengths of both convolutional and recurrent neural network components, the proposed models are able to effectively capture spatial features as well as model the temporal dynamics and contextual relationships present in handwritten Arabic text. Experiments conducted on the AHCD and Hijjaa benchmark datasets show that the CNN-Bi-GRU framework achieved state-of-the-art accuracy rates of 97.05% and 91.78% respectively, outperforming previous deep learning-based methods. These results demonstrate the significant performance gains that can be achieved by integrating specialized temporal modeling and contextual representation capabilities into the handwriting recognition pipeline, without the need for explicit segmentation. The findings of this research represent a crucial advancement in the continued development of sophisticated and precise deep learning systems for Arabic handwriting recognition, with broad applications across domains that rely on efficient text extraction from handwritten documents.

Deep transformer-based architecture for the recognition of mathematical equations from real-world math problems

Identifying mathematical equations from real-world math problems presents a unique and challenging task within the field of Natural Language Processing (NLP). It has a wide range of applications in various areas, such as academics, digital content design, and the development of automatic or interactive learning systems. However, the accurate understanding of these equations still needs to be enhanced due to the intrinsic complexity of mathematical symbols and various structural formats. Additionally, the unique syntax, diverse symbols, and complex structure of mathematical equations present significant obstacles that traditional NLP methods and Optical Character Recognition (OCR) systems need help to overcome. In this research, we utilize deep transformer architecture to recognize mathematical equations, and we have utilized our novel dataset with 3433 distinct observations. This dataset, which we have collected to include a diverse range of mathematical equations, is used to predict six (z=x+y, z=x−y, z=xy, z=x/y, y=x! and y=x) basic mathematical equations. We applied different transformer-based architectures, such as BERT, ELECTRA, XLNet, RoBERTa, and DistilBERT, and BERT performs best with 99.80% accuracy. To the best of our knowledge, this is the first NLP work in any language where we recognize the equation from the mathematical text.

Influence Of Principals’ Supportive Leadership Style On Tutors’ Support For Implementation Of Performance Appraisal Of Tutors In Public Primary Teachers Training Colleges In Kenya

The study objectives sought; to establish the influence of principals’ supportive leadership style on tutors’ support for implementation of performance appraisal in public primary teachers training colleges in Kenya. The study employed descriptive research design. The target population comprised of 840 tutors and 21 principals drawn from the 21 public teachers training colleges in Kenya as of 2021. Stratified proportionate random sampling was used to select 271 tutors in teachers’ training colleges in Kenya. Census was used to select all the 21 principals of training colleges in Kenya. Primary data were collected using questionnaire for tutors and interview guide for the 21 Principals. The study findings demonstrated that supportive leadership style of principals showed a significant positive effect on tutors' support for performance appraisal (β = 0.386, p < 0.001). The study concluded that principals' leadership styles supportive is critical determinants in enhancing tutors' engagement and commitment to the appraisal process. There is need for the college management to provide the necessary teaching and training materials to support the effective implementation of performance appraisal programmes. There is also need for tutors to present their past output so that it can inform supportive principal what is necessary for a better working environment. The study also recommends that the MoE, alongside college management, prioritize the creation of policies that promote a supportive and safe working environment. This could include mentorship programs for tutors, flexible scheduling, and recognition and reward systems to acknowledge tutors’ professional growth and accomplishments

Enhanced Vehicle Recognition and Collision Alert System Using Yolo+DeepSORT Algorithm

Abstract This paper sets out to develop and deploy a deep learning-driven vehicle collision avoidance warning system. Initially, it delves into the foundational principles of traditional collision avoidance warning systems, alongside key safety driving concepts. It then elaborates on leveraging deep learning technology to enhance the precision and reliability of target tracking. This includes detailed discussions on techniques such as appearance feature extraction, Kalman filters, the Hungarian algorithm, and more. In the technical section, the paper outlines the specific implementation of utilizing the YOLO algorithm for target detection and the DeepSORT algorithm for target tracking within the system architecture. The experimental section showcases the system’s performance and stability through formatted experimental results, followed by a comprehensive analysis of the data obtained. Concluding, the paper summarizes its research findings, underscoring the significance and feasibility of a deep learning-based vehicle collision avoidance warning system.

SERKs serve as co-receptors for SYR1 to trigger systemin-mediated defense responses in tomato.

Systemin, the first peptide hormone identified in plants, was initially isolated from tomato (Solanum lycopersicum) leaves. Systemin mediates local and systemic wound-induced defense responses in plants, conferring resistance to necrotrophic fungi and herbivorous insects. Systemin is recognized by the leucine-rich-repeat receptor-like kinase (LRR-RLK) receptor SYSTEMIN RECEPTOR1 (SYR1), but how the systemin recognition signal is transduced to intracellular signaling pathways to trigger defense responses is poorly understood. Here, we demonstrate that SERK family LRR-RLKs function as co-receptors for SYR1 to mediate systemin signal transduction in tomato. By using chemical genetic approaches coupled with engineered receptors, we revealed that the association of the cytoplasmic kinase domains of SYR1 with SERKs leads to their mutual trans-phosphorylation and the activation of SYR1, which in turn induces a wide range of defense responses. Systemin stimulates the association between SYR1 and all tomato SERKs (SlSERK1, SlSERK3A, and SlSERK3B). The resulting SYR1-SlSERK heteromeric complexes trigger the phosphorylation of TOMATO PROTEIN KINASE 1B (TPK1b), a receptor-like cytoplasmic kinase that positively regulates systemin responses. Additionally, upon association with SYR1, SlSERKs are cleaved by the Pseudomonas syringae effector HopB1, further supporting the finding that SlSERKs are activated by systemin-bound SYR1. Finally, genetic analysis using Slserk mutants showed that SlSERKs are essential for systemin-mediated defense responses. Collectively, these findings demonstrate that the systemin-mediated association of SYR1 and SlSERKs activates defense responses against herbivorous insects.

Enhanced sensitivity and broadband response in porous triple periodic minimal surface piezoresistive sensors for telemedicine applications

Wearable pressure sensors with outstanding performance have garnered significant attention in fields such as telemedicine. However, the development of sensors that achieve both adjustable high sensitivity and a wide response range remains a formidable challenge. To address these issues, we propose the development of a flexible piezoresistive sensor featuring simultaneous macro- and microstructures. This is achieved by constructing a complex macro-topological skeleton using Fused Deposition Molding (FDM) technology and integrating supercritical carbon dioxide (ScCO2) foaming technology for microscopic pore creation. The unique conductive network structure, coupled with the Triple Periodic Minimal Surface (TPMS) metamaterial framework and microscopic pore architecture, endows the piezoresistive sensors with exceptional sensing performance, including high sensitivity (12.13 MPa−1), a broad response range (exceeding 30 MPa), and long-term fatigue resistance (over 26,000 s cycles). In addition, the sensors are shielded from electromagnetic interference, with a maximum EMI shielding effectiveness of 53.2 dB for the TMFM with a thickness of 2.3 mm. The effects of varying porosities and foaming pressures on sensitivity were analyzed using finite element simulations. Owing to these remarkable features, multiple foot movement recognition was successfully demonstrated with volunteer participants by assembling sensing unit array modules. Based on this, an intelligent fall warning recognition system is developed using machine learning. This system accurately distinguishes four types of activity behaviors—standing, walking, spraining foot, and falling—with accuracy rates of 98.9 %, 97.1 %, 98.9 %, and 95.1 %, respectively. The system automatically generates an alarm pattern for detected fall behaviors. These findings suggest that the combination of TPMS skeleton and micropores structure offers a promising strategy for developing sensors with adjustable sensitivity and wide detection range for wearable devices. This technology holds significant potential for applications in telemedicine assistance, particularly for the elderly and other vulnerable populations.

AI-powered precursor quantification in atmospheric pressure plasma jet thin film deposition via optical emission spectroscopy

Abstract This study explores the feasibility of using Optical Emission Spectroscopy (OES) for in situ monitoring of Atmospheric Pressure Plasma Jet systems in the deposition of thin films. We identify process parameters to control film properties by machine learning for data analysis. In experiments, the depth of the carrier gas inlet pipe (pipe depth) is a crucial controllable variable that directly affects the amount of precursor, influencing the film’s thickness, sheet resistance, and resistivity. We collected 96 000 spectra while preparing 12 film samples, subsequently measured the properties of the samples, and analyzed the spectral data using Principal Component Analysis (PCA) and seven supervised machine learning models. A high correlation was found between spectral features and film thickness. We divided the spectral data in a single process based on processing time into the first third (F-third) and the last third (L-third). Using the F-third data, the PCA plot clearly indicated a significant difference between the two pipe depths, achieving a mean recognition accuracy of 95.1% with machine learning models. In contrast, using the L-third data, the PCA plot showed a high degree of overlap between the two pipe depths, resulting in a considerable decline in recognition performance. Overall, it is challenging to distinguish the spectra visually due to variations in precursor amounts and dynamic fluctuations in the OES signals, even after averaging. Nonetheless, through the successful application of machine learning, we demonstrated an effective spectral recognition system for monitoring pipe depth, which aids in the timely control of film properties.

A Deep Learning Application of Capsule Endoscopic Gastric Structure Recognition Based on a Transformer Model.

Gastric structure recognition systems have become increasingly necessary for the accurate diagnosis of gastric lesions in capsule endoscopy. Deep learning, especially using transformer models, has shown great potential in the recognition of gastrointestinal (GI) images according to self-attention. This study aims to establish an identification model of capsule endoscopy gastric structures to improve the clinical applicability of deep learning to endoscopic image recognition. A total of 3343 wireless capsule endoscopy videos collected at Nanfang Hospital between 2011 and 2021 were used for unsupervised pretraining, while 2433 were for training and 118 were for validation. Fifteen upper GI structures were selected for quantifying the examination quality. We also conducted a comparison of the classification performance between the artificial intelligence model and endoscopists by the accuracy, sensitivity, specificity, and positive and negative predictive values. The transformer-based AI model reached a relatively high level of diagnostic accuracy in gastric structure recognition. Regarding the performance of identifying 15 upper GI structures, the AI model achieved a macroaverage accuracy of 99.6% (95% CI: 99.5-99.7), a macroaverage sensitivity of 96.4% (95% CI: 95.3-97.5), and a macroaverage specificity of 99.8% (95% CI: 99.7-99.9) and achieved a high level of interobserver agreement with endoscopists. The transformer-based AI model can accurately evaluate the gastric structure information of capsule endoscopy with the same performance as that of endoscopists, which will provide tremendous help for doctors in making a diagnosis from a large number of images and improve the efficiency of examination.

Dual Attention Adversarial Attacks With Limited Perturbations.

The construction of undetectable adversarial examples with few perturbances remains a difficult problem in adversarial attacks. At present, most solutions use the standard gradient optimization algorithm to build adversarial examples by applying global perturbations to benign samples and then launch attacks on the targets (e.g., face recognition systems). However, when the perturbance size is limited, the performance of these approaches suffers substantially. The content of crucial places in an image, on the other hand, will impact the final prediction; if these areas can be investigated and limited perturbances introduced, an acceptable adversarial example will be constructed. Based on the foregoing research, this article offers a dual attention adversarial network (DAAN) to produce adversarial examples with limited perturbations. DAAN initially searches for effective areas in an input image using the spatial attention network and channel attention network, and then creates space and channel weights. Following that, these weights direct an encoder and a decoder to generate effective perturbation, which is then combined with the input to produce an adversarial example. Finally, the discriminator determines if the created adversarial examples are true or false, and the attacked model is utilized to determine whether the generated samples fit the attack targets. Extensive studies on various datasets show that DAAN not only delivers the best attack performance across all comparison algorithms with few perturbations, but it can also significantly improve the defensiveness of the attacked models.

Transfer learning with YOLOV8 for real-time recognition system of American Sign Language Alphabet

Sign language serves as a sophisticated means of communication vital to individuals who are deaf or hard of hearing, relying on hand movements, facial expressions, and body language to convey nuanced meaning. American Sign Language (ASL) exemplifies this linguistic complexity with its distinct grammar and syntax. The advancement of real-time ASL gesture recognition has explored diverse methodologies, including motion sensors and computer vision techniques. This study specifically addresses the recognition of ASL alphabet gestures using computer vision through Mediapipe for hand movement tracking and YOLOv8 for training the deep learning model. The model achieved notable performance metrics: precision of 98%, recall rate of 98%, F1 score of 99%, mean Average Precision (mAP) of 98%, and mAP50-95 of 93%, underscoring its exceptional accuracy and sturdy capabilities.

Bimodal biometric recognition system using Convolutional Neural Networks and fusion of deep audiovisual feature vectors

In recent years, interest has grown in the use biometric systems for identity authentication tasks in digital services, forensic and security applications. A unimodal system (employing a single biometric trait) with high performance is still vulnerable to falsification attacks such as spoofing. For this reason, research on multimodal biometrics (employing various biometric traits) has increased to reinforce security, increase recognition performance, and make false identity authentication more difficult. In this paper, we propose a bimodal system that combines speech and face modalities by concatenating their feature vectors, these vectors are extracted from two convolutional neural networks (CNN) and used for identity verification. The performance of unimodal CNNs was evaluated individually and compared to the bimodal system of concatenated vectors. A data augmentation scheme is used for both modalities to evaluate different operation conditions. Results were measured in terms of Equal Error Rate (EER).

Ensem-DeepHAR: Identification of human activity in smart environments using ensemble of deep learning methods and motion sensor data

Recognizing human activity plays a crucial role in many applications such as medical care services in smart healthcare environments. Inertial or motion sensors can measure physiognomies such as acceleration and angular velocity of body movement while performing the activities and we can use them to learn the models capable of activity recognition. Over the past decades, many state-of-the-art activity recognition systems have been developed but there is still room to improve. In this paper, we have proposed a novel approach to identify human activity from motion sensor data by employing an enormous analysis of sensor data. Based on data analysis, we yielded quality data by preprocessing using a preprocessing chain for human activity recognition (PC-HAR) which also includes the Synthetic Minority Over-sampling Technique to balance the data of the dataset. As a recognition model, we proposed an ensemble of three different deep learning algorithms, namely, modified DeepConvLSTM, modified InceptionTime, and modified ResNet which is named ‘Ensem-DeepHAR’. The outcome of the proposed model is carried out by stacking predictions from each of the mentioned models and then a Random Forest as a meta-model uses those predictions to recognize the final activity. We evaluated our method on both person-dependent and person-independent cases and achieved 99.31 %, 99.08 %, and 97.52 % accuracies for the former case and 97.95 %, 98.11 %, and 99.51 % accuracies for the latter case using three common benchmark datasets: WISDM_ar_v1.1, PAMAP2, and UCI-HAR respectively. The various performance metrics and measures of experimental results establish the supremacy of the proposed model over the state-of-the-arts.