High Recognition Accuracy Research Articles

Extending Dynamic Blind Interaction: Microgap Discharge-Induced Flexible Virtual Electrotactile Braille.

Braille is an essential implement for the blind to communicate with outside, but traditional Braille is limited to a paper-based format that cannot directly provide real-time word information. In this work, a flexible virtual electrotactile Braille is proposed that can benefit the blind from blocked interaction. The Braille interface, S-shaped wires and a sphere electrode with a textile fingerstall integrated by silicone, offers flexibility and simultaneously generates the microgap through textile cracks, which achieves virtual electrotactile sensation by electrostatic discharge. Powered by a high-voltage triboelectric generator of 10.2 kV designed through the charge accumulation and induction strategy, the electrotactile stimulation is realized with a microgap discharge of only 40 μA current induced on the finger. A dynamic electrotactile Braille is finally assembled, controlled by a programmable relay array. The strategies of short circuit and voice reminder are employed, so that the recognition of dynamic Braille letters is realized with spatiotemporal electrotactile stimulation and high recognition accuracy. This virtual electrotactile Braille brings convenience for the blind to access the information world and illustrates its applications to promote virtual electrotactility in this special community.

Biodegradable, stretchable, and high-performance triboelectric nanogenerators through interfacial polarization in bilayer structure

The increasing demand for wearable electronics has led to the development of triboelectric nanogenerators (TENGs) as a promising energy harvesting and sensing technology. However, conventional TENGs often utilize non-biodegradable materials, contributing to environmental pollution. In this work, we present a stretchable and biodegradable TENG based on hydroxyethyl cellulose (HEC) and gelatin (HG-TENG). The HG-TENG features a bilayered structure, where the large difference in their relative permittivity between HEC and gelatin induces interfacial polarization, effectively mitigating charge recombination and enhancing triboelectric performance. The optimized HG-TENG achieves an open-circuit voltage (Voc) of 93 V, a maximum power density of 57.8 µW/cm2, and can power 38 blue light-emitting diodes. The device exhibits a stretchability of 150 % and biodegrades within 3 hours in phosphate-buffered saline. Furthermore, we demonstrate the application of the HG-TENG as a wearable sensor by modifying it with trichloro(1H, 1H, 2H, 2H-perfluorooctyl)silane (FOTS) (FHG-TENG). The FHG-TENG-based smart glove, integrated with machine learning algorithms, enables real-time monitoring of blood pressure waveforms and finger motions, showcasing its potential for human-machine interfaces. The smart glove, equipped with five FHG-TENGs on the proximal interphalangeal joints of each finger, detects diverse finger gestures and generates voltage signals that control a robotic hand in real-time, demonstrating effective human-machine interaction through synchronized motion. Moreover, the smart glove achieves a high recognition accuracy of 96.15 % for 10 different hand sign languages.

Direct implementation of AI-Based Facial Recognition for ITSI students

The development of artificial intelligence (AI)-based facial recognition technology has become a significant research topic in the field of computing and security. At the Indonesian Palm Oil Institute (ITSI), AI-based facial recognition is introduced to students to improve their skills in developing AI-based applications. This study aims to implement and test a facial recognition system using a Python program by utilizing a dataset generated independently. This research method involves several stages, namely collecting ITSI students' facial data, data processing, creating a facial recognition model using a machine learning algorithm, and evaluating model performance. The dataset used was developed through a live shooting session involving active student participation. The facial recognition model was trained using a convolutional neural network (CNN) algorithm that was optimized to improve accuracy. The results of the study showed that the developed model was able to achieve high facial recognition accuracy, with an average accuracy rate of 92%. The discussion includes an analysis of factors that affect accuracy, such as variations in lighting and shooting angles, as well as the potential use of this technology in a campus environment, including for attendance and security purposes. The conclusion of this study shows that the implementation of AI-based facial recognition can be effectively applied in an academic environment, as well as providing students with practical experience in developing and testing AI applications. This study also opens up opportunities for further research on improving the performance of facial recognition systems and their application in various real-world scenarios.

Modulation Format Recognition Scheme Based on Discriminant Network in Coherent Optical Communication System

In this paper, we skillfully utilize the discriminative ability of the discriminator to construct a conditional generative adversarial network, and propose a scheme that uses few symbols to achieve high accuracy recognition of modulation formats under low signal-to-noise ratio conditions in coherent optical communication. In the one thousand kilometres G.654E optical fiber transmission system, transmission experiments are conducted on the PDM-QPSK/-8PSK/-16QAM/-32QAM/-64QAM modulation format at 8G/16G/32G baud rates, and the signal-to-noise ratio parameters are traversed under experimental conditions. As a key technology in the next-generation elastic optical networks, the modulation format recognition scheme proposed in this paper achieves 100% recognition of the above five modulation formats without distinguishing signal transmission rates. The optical signal-to-noise ratio thresholds required to achieve 100% recognition accuracy are 12.4 dB, 14.3 dB, 15.4 dB, 16.2 dB, and 17.3 dB, respectively.

Accelerable adaptive cepstrum and L2-Dual Net for acoustic emission-based quality monitoring in laser shock peening

Acoustic emission monitoring in laser shock peening facilitates real-time detection of potential quality issues arising from variations in industrial parameters, enabling iterative optimization of the manufacturing process through material behavior analysis. However, existing research still lacks a comprehensive understanding of the time-varying time-frequency characteristics in dynamic acoustic emission and efficient corresponding models. Therefore, this study proposes an innovative monitoring approach that integrates accelerable adaptive cepstrum (AAC) and L2-Dual Net. Specifically, AAC first employs variable frames and filters to map time-varying features in the signal, and then obtains representative frame length distributions and filter weights for different operating conditions based on statistical information. AAC not only unveils time-varying features in signals but also boasts an efficient computational process. L2-Dual Net is a novel quality assessment model with robust feature extraction and local spatial feature interactions. The incorporation of L2 norm equips the model with robust interference immunity, while the dual spatial attention mechanism helps the model to interact with spatial features exhibiting different time-frequencies. Variable process parameter experiments for aluminum alloy 7075 and titanium alloy TC4 were conducted to validate the reliability of the proposed method. Results demonstrate that AAC showcases optimal computational efficiency and higher feature resolution. When compared with state-of-the-art network architectures, L2-Dual Net exhibits superior information flow, along with higher recognition accuracy and robustness. Moreover, various variants of L2-Dual Net are explored and the code is accessible at https://github.com/Qinr1026/L2-Dual-Net. The proposed method holds promising potential for application in other areas of acoustic emission monitoring.

Reconfigurable Five-In-One Carbon Nanotube Optoelectronic Transistor for Intelligent Computing and Communication.

Advances in artificial general intelligence (AGI) necessitate the integration of diverse functionalities to address complex tasks. Carbon nanotubes (CNTs), with their unique physical properties, have broad applications in emerging research fields, providing a foundation for next-generation devices that could overcome the limits of Moore's Law. This work demonstrates a novel intelligent device that integrates five functions─sensors, memory, neuromorphic computing, logic, and communication─using CNT field-effect transistors (CNFETs) compatible with CMOS processes. Through passivation and annealing techniques, we have significantly enhanced the optoelectronic performance of CNFETs, leading to the development of multifunctional optoelectronic synaptic transistors. These optimized CNFETs enable dual-mode weight-tunable synaptic functions, including long-term plasticity and multilevel storage. Additionally, a CNT-based neural network has achieved high recognition accuracy on the MNIST data set, showcasing the potential of in-memory computing. This research also innovates by integrating logic functions with optoelectronic communication capabilities, paving the way for next-generation intelligent computing and communication integrated systems.

Pattern recognition using spiking antiferromagnetic neurons

Spintronic devices offer a promising avenue for the development of nanoscale, energy-efficient artificial neurons for neuromorphic computing. It has previously been shown that with antiferromagnetic (AFM) oscillators, ultra-fast spiking artificial neurons can be made that mimic many unique features of biological neurons. In this work, we train an artificial neural network of AFM neurons to perform pattern recognition. A simple machine learning algorithm called spike pattern association neuron (SPAN), which relies on the temporal position of neuron spikes, is used during training. In under a microsecond of physical time, the AFM neural network is trained to recognize symbols composed from a grid by producing a spike within a specified time window. We further achieve multi-symbol recognition with the addition of an output layer to suppress undesirable spikes. Through the utilization of AFM neurons and the SPAN algorithm, we create a neural network capable of high-accuracy recognition with overall power consumption on the order of picojoules.

Object Motion Detection Enabled by Reconfigurable Neuromorphic Vision Sensor under Ferroelectric Modulation.

Increasing the demand for object motion detection (OMD) requires shifts of reducing redundancy, heightened power efficiency, and precise programming capabilities to ensure consistency and accuracy. Drawing inspiration from object motion-sensitive ganglion cells, we propose an OMD vision sensor with a simple device structure of a WSe2 homojunction modulated by a ferroelectric copolymer. Under optical mode and intermediate ferroelectric modulation, the vision sensor can generate progressive and bidirectional photocurrents with discrete multistates under zero power consumption. This design enables reconfigurable devices to emulate long-term potentiation and depression for synaptic weights updating, which exhibit 82 states (more than 6 bits) with a uniform step of 6 pA. Such OMD devices also demonstrate nonvolatility, reversibility, symmetry, and ultrahigh linearity, achieving a fitted R2 of 0.999 and nonlinearity values of 0.01/-0.01. Thus, a vision sensor could implement motion detection by sensing only dynamic information based on the brightness difference between frames, while eliminating redundant data from static scenes. Additionally, the neural network utilizing a linear result can recognize the essential moving information with a high recognition accuracy of 96.8%. We also present the scalable potential via a uniform 3 × 3 neuromorphic vision sensor array. Our work offers a platform to achieve motion detection based on controllable and energy-efficient ferroelectric programmability.

An open-access database of video stimuli for action observation research in neuroimaging settings: psychometric evaluation and motion characterization.

Video presentation has become ubiquitous in paradigms investigating the neural and behavioral responses to observed actions. In spite of the great interest in uncovering the processing of observed bodily movements and actions in neuroscience and cognitive science, at present, no standardized set of video stimuli for action observation research in neuroimaging settings exists. To facilitate future action observation research, we developed an open-access database of 135 high-definition videos of a male actor performing object-oriented actions. Actions from 3 categories: kinematically natural and goal-intact (Normal), kinematically unnatural and goal-intact (How), or kinematically natural and goal-violating (What), directed toward 15 different objects were filmed from 3 angles. Psychometric evaluation of the database revealed high video recognition accuracy (Mean accuracy = 88.61 %) and substantial inter-rater agreement (Fleiss' Kappa = 0.702), establishing excellent validity and reliability. Videos' exact timing of motion onset was identified using a custom motion detection frame-differencing procedure. Based on its outcome, the videos were edited to assure that motion begins at the second frame of each video. The videos' timing of category recognition was also identified using a novel behavioral up-down staircase procedure. The identified timings can be incorporated in future experimental designs to counteract jittered stimulus onsets, thus vastly improving the sensitivity of neuroimaging experiments. All videos, their psychometric evaluations, and the timing of their frame of category recognition, as well as our custom programs for performing these evaluations on our, or on other similar video databases, are available at the Open Science Framework (https://osf.io/zexc4/).

AI-enhanced backpack with double frequency-up conversion vibration energy converter for motion recognition and extended battery life

With growing interest in physical activities, wearable device sales have been increasing each year due to their help in tracking movement. However, keeping these devices running for a long time is still a key challenge. This paper presents an AI-enhanced backpack equipped with a hybrid vibration energy converter that integrates piezoelectric (PE) and electromagnetic (EM) technologies, featuring a double frequency-up conversion (FUC) mechanism. The PE unit enables efficient motion recognition, while the EM unit extends battery life via kinetic energy harvesting. The double FUC mechanism boosts human motion signal strength and resolution by converting sub-Hertz frequencies to hundreds of Hertz. This makes the device particularly effective in environments with performance constraints and interference. By adopting deep learning techniques, the system maintains a high motion recognition accuracy of 97.33 % even under low data sampling rates and significant noise interference. The EM unit effectively harvests kinetic energy, generating an average output power of 4.5 mW during activities such as running at 4.5 Hz, despite of the device’s compact size. Additionally, a power management circuit with human motion-stimulated switch optimizes battery life by managing the system’s sleep and active states. This innovative energy management approach ensures that during human activity, the system conserves energy by remaining in sleep state for at least 30 % of the time, significantly extending battery life and enhancing the backpack’s suitability for outdoor applications.

Ultrasensitive and wide-range MXene/PDMS piezoresistive sensors inspired by rose petals

Flexible pressure sensors are crucial in a variety of applications, including artificial intelligence (AI), the Internet of Things (IoT), and electronic skin (e-Skin), due to their ability to convert mechanical stimuli into electrical signals. However, achieving both high sensitivity and a wide detection range simultaneously remains challenging, often limiting their practical applications. To address this, we designed a three-dimensional contact interface with Ti3C2 MXene/polydimethylsiloxane (MP) inspired by rose petals, which allows for significant strain and resistance changes across a broad range of stress levels. The innovative design significantly extends the pressure detection range (0.033–535 kPa), enhances pressure sensitivity (0.432 kPa−1), offers a fast response/recovery times (57/4 ms), and ensures exceptional long-term durability (up to 2950 cycles). The structured MP sensor demonstrates the capability to detect human physiological movements, distinguish various voices and map real-time pressure distributions. Notably, it achieves real-time tracking of handwriting processes with high recognition accuracy through a machine learning module. The work offers a new strategy and inspiration for designing and manufacturing innovative structured pressure sensors with reliable performance, paving the way for their application in next-generation wearable and portable electronics.

The Development of Bi-LSTM Based on Fault Diagnosis Scheme in MVDC System

Diagnosing faults is crucial for ensuring the safety and reliability of medium-voltage direct current (MVDC) systems. In this study, we propose a bidirectional long short-term memory (Bi-LSTM)-based fault diagnosis scheme for the accurate classification of faults occurring in MVDC systems. First, to ensure stability in case a fault occurs, we modeled an MVDC system that included a resistor-based fault current limiter (R-FCL) and a direct current circuit breaker (DCCB). A discrete wavelet transform (DWT) extracted the transient voltages and currents measured using DC lines and AC grids in the frequency–time domain. Based on the digital signal normalized by the DWT, using the measurement data, the Bi-LSTM algorithm was used to classify and learn the types and locations of faults, such as DC line (PTP, P-PTG, and N-PTG) and internal inverter faults. The effectiveness of the proposed fault diagnosis scheme was validated through comparative analysis within the four-terminal MVDC system, demonstrating superior accuracy and a faster diagnosis time compared to those of the existing schemes that utilize other AI algorithms, such as the CNN and LSTM. According to the test results, the proposed fault diagnosis scheme detects MVDC faults and shows a high recognition accuracy of 97.7%. Additionally, when applying the Bi-LSTM-based fault diagnosis scheme, it was confirmed that not only the training diagnosis time (TraDT) but also the average diagnosis time (AvgDT) were 0.03 ms and 0.05 ms faster than LSTM and CNN, respectively. The results validate the superior fault clarification and fast diagnosis performance of the proposed scheme over those of the other methods.

Dual Redox-active Covalent Organic Framework-based Memristors for Highly-efficient Neuromorphic Computing.

Organic memristors based on covalent organic frameworks (COFs) exhibit significant potential for future neuromorphic computing applications. The preparation of high-quality COF nanosheets through appropriate structural design and building block selection is critical for the enhancement of memristor performance. In this study, a novel room-temperature single-phase method was used to synthesize Ta-Cu3 COF, which contains two redox-active units: trinuclear copper and triphenylamine. The resultant COF nanosheets were dispersed through acid-assisted exfoliation and subsequently spin-coated to fabricate a high-quality COF film on an indium tin oxide (ITO) substrate. The synergistic effect of the dual redox-active centers in the COF film, combined with its distinct crystallinity, significantly reduces the redox energy barrier, enabling the efficient modulation of 128 non-volatile conductive states in the Al/Ta-Cu3 COF/ITO memristor. Utilizing a convolutional neural network (CNN) based on these 128 conductance states, image recognition for ten representative campus landmarks was successfully executed, achieving a high recognition accuracy of 95.13 % after 25 training epochs. Compared to devices based on binary conductance states, the memristor with 128 conductance states exhibits a 45.56 % improvement in recognition accuracy and significantly enhances the efficiency of neuromorphic computing.

Evaporated Copper‐Based Perovskite Dynamic Memristors for Reservoir Computing Systems

AbstractDynamic memristors are considered as the optimal hardware devices for reservoir computing (RC) enabled by their nonlinear conductance variations. This significantly reduces the extensive training workload typically required by traditional neural networks. Lead halide perovskites, with their tunable band structure and active ion migration properties, have emerged as highly promising materials for developing dynamic memristors. However, large‐scale and consistently stable production remains a challenge for perovskite functional films, while lead elements' toxicity and environmental impact also partly restrict their practical device utilization. In this work, lead‐free copper‐based perovskite (i.e., CsCu2I3) films are prepared by thermal evaporation for constructing dynamic memristors. The effective conductivity modulation of CsCu2I3‐based memristor can be utilized in artificial neural networks, achieving a high handwritten digit recognition accuracy of 91.2%. In addition, the RC system is also constructed based on the dynamic behavior of the devices, by which a letter recognition accuracy of 98.2% with simple training is achieved. This technology provides a feasible pathway to construct copper‐based perovskite dynamic memristors for future neural network information processing.

An ultra lightweight neural network for automatic modulation classification in drone communications

Unmanned aerial vehicle (UAV)-assisted communication based on automatic modulation classification (AMC) technology is considered an effective solution to improve the transmission efficiency of wireless communication systems, as it can adaptively select the most suitable modulation method according to the current communication environment. However, many existing deep learning (DL)-based AMC methods cannot be directly applied to UAV platform with limited computing power and storage space, because of the contradiction between accuracy and efficiency. This paper mainly studies the lightweight of DL-based AMC networks to improve adaptability in resource-constrained scenarios. To address this challenge, we propose an ultra-lightweight neural network (ULNN). This network incorporates a lightweight convolutional structure, attention mechanism, and cross-channel feature fusion technique. Additionally, we introduce data augmentation (DA) based on signal phase offsets during the model training process, aimed at improving the model’s generalization ability and preventing overfitting. Through experimental validation on the public dataset RML2016.10 A, the ULNN we proposed achieves an average precision of 62.83% with only 8815 parameters and reaches a peak classification accuracy of 92.11% at SNR = 10 dB. The experimental results show that ULNN can achieve high recognition accuracy while keeping the model lightweight, and is suitable for UAV platform with limited resources.

Artificial Intelligence in Dermatopathology: a systematic review.

Medical research, driven by advancing technologies like Artificial Intelligence (AI), is transforming healthcare. Dermatology, known for its visual nature, benefits from AI, especially in dermatopathology with digitized slides. This review explores into AI's role, challenges, opportunities, and future potential in enhancing dermatopathological diagnosis and care. Adhering to PRISMA and Cochrane Handbook standards, this systematic review explored AI's function in dermatopathology. It employed an interdisciplinary method, encompassing diverse study types and comprehensive database searches. Inclusion criteria encompassed peer-reviewed articles from 2000 to 2023, with a focus on practical AI use in dermatopathology. Numerous studies have investigated AI's potential in dermatopathology. We reviewed 112 papers. Notable applications include AI classifying histopathological images of nevi and melanomas, although challenges exist regarding subtype differentiation and generalizability. AI achieved high accuracy in melanoma recognition from formalin-fixed paraffin-embedded samples but faced limitations due to small datasets. Deep learning algorithms showed diagnostic accuracy for specific skin conditions, but challenges persisted, such as small sample sizes and the need for prospective validation. This systematic review underscores AI's potential in enhancing dermatopathology for better diagnosis and patient care. Addressing challenges like limited datasets and potential biases is essential. Future directions involve expanding datasets, conducting validation studies, promoting interdisciplinary collaboration, and creating patient-centred AI tools to enhance dermatopathology's accuracy, accessibility, and patient-focused care.

A Deformable and Multi-Scale Network with Self-Attentive Feature Fusion for SAR Ship Classification

The identification of ships in Synthetic Aperture Radar (SAR) imagery is critical for effective maritime surveillance. The advent of deep learning has significantly improved the accuracy of SAR ship classification and recognition. However, distinguishing features between different ship categories in SAR images remains a challenge, particularly as the number of categories increases. The key to achieving high recognition accuracy lies in effectively extracting and utilizing discriminative features. To address this, we propose DCN-MSFF-TR, a novel recognition model inspired by the Transformer encoder–decoder architecture. Our approach integrates a deformable convolutional module (DCN) within the backbone network to enhance feature extraction. Additionally, we introduce multi-scale self-attention processing from the Transformer into the feature hierarchy and fuse these representations at appropriate levels using a feature pyramid strategy. This enables each layer to leverage both its own information and synthesized features from other layers, enhancing feature representation. Extensive evaluations on the OpenSARShip-3-Complex and OpenSARShip-6-Complex datasets demonstrate the effectiveness of our method. DCN-MSFF-TR achieves average recognition accuracies of 78.1% and 66.7% on the three-class and six-class datasets, respectively, outperforming existing recognition models and showcasing its superior capability in accurately identifying ship categories in SAR images.

Face Recognition Attendance Management System using LBPH and Haar Cascade

This research presents the implementation and design of a facial recognition attendance management system by using Local Binary Pattern Histogram (LBPH) and Haar Cascade machine learning algorithms. This system aims to automate the attendance process, providing an efficient and accurate alternative to traditional methods. The Haar Cascade algorithm is employed for face detection due to its rapid processing and high detection rate, while the LBPH algorithm is utilized for face recognition because it is simple and effective in handling changes in facial expressions, lighting, occlusions, distance from the camera, and camera resolution. The integration of these algorithms results in a robust system capable of real-time attendance tracking. Experimental results demonstrate the system's high accuracy and reliability in face detection and recognition under different conditions of lighting, distance from the camera, face expressions, occlusions, and camera resolution, making it suitable for deployment in educational institutions and corporate environments.

Improved Pedestrian Vehicle Detection for Small Objects Based on Attention Mechanism

Abstract This study aims to solve the low detection accuracy and susceptibility to false detection and omission in pedestrian and vehicle detection by proposing an improved YOLOv5s algorithm. Firstly, a small target detection module is added to better acquire and determine the information of pedestrians from long-range vehicles. Secondly, the multi-scale channel attention CBAM attention module is added, and the dual attention mechanism is not only flexible and convenient, but also improves the computational efficiency. Finally, the MPDIoU loss function based on minimum point distance is introduced to replace the original GIoU loss function, and this change not only enhances the regression accuracy of the model. At the same time, the convergence speed of the model is accelerated. KITTI data set was used for experiments, and the experimental results showed that the average accuracy of the model trained by the improved YOLOv5s algorithm on the data set reached 84.9%, which was 3.7% higher than that of the original YOLOv5s algorithm. It is verified that the model is suitable for high accuracy of pedestrian and vehicle recognition in complex environments, and has high value for promotion.

Morphological dictionary learning based sparse classification for small electric motor state recognition under unbalanced samples

Accurately recognizing sound states in the production line of small electric motors is of great importance for manufacturers to carry out quick repairs and ensure high quality deliveries. Since the number of normal samples is much larger than the number of abnormal samples in practice, resulting in unbalanced data, which poses huge challenges to traditional detection methods. To overcome these difficulties, this study presents a morphological dictionary learning-based sparse classification (MDL-SC) combined with audio data augmentation method for small electric motor state recognition under unbalanced samples. Firstly, audio data augmentation methods such as adding background noise, pitch shifting, time stretching and combined augmentation are investigated for augmenting the number and diversity of samples. Secondly, morphological dictionary learning is proposed for characterizing transient sounds of small electric motors and enhancing the discriminative feature learning capability of the dictionary. Finally, the minimum reconstruction error strategy is relied upon to establish automatic recognition of small electric motor states. Three small motor datasets with unbalanced ratios are established in the experiments to verify the effectiveness of the proposed MDL-SC, which has higher recognition accuracy under unbalanced conditions compared with traditional dictionary learning based sparse classification (DL-SC), k-nearest neighbors, support vector machines and convolutional neural networks. This study can provide some theoretical implications for the later development of online detection of small electric motors or other types of electric motors.