Search

Seepage failure of levees can cause landslides and other hazards. Three-dimensional (3D) laser-scanning technology has become a new method for collecting levee hazard data. In this study, the 3D characteristics of landslide point clouds were investigated through systematic indoor model tests, and a feature recognition algorithm applicable to levee landslides was proposed. The major outcomes of this study are as follows: 1) the formulation of an adaptive random sampling boundary extraction (A-R-B) algorithm, which integrates random sample consensus plane segmentation, adaptive distance threshold calculation, and boundary extraction for levee landslide disaster recognition; 2) through feasibility analyses and accuracy tests of the A-R-B algorithm, this study demonstrated the capacity of the proposed method to accurately recognise the features of levee landslides, with a relative accuracy of 1 cm and an absolute accuracy of 3.5 cm in the extraction process; 3) the testing of the A-R-B algorithm and optimal parameters for the recognised levee landslide features using the point clouds obtained from laboratory models. [ABSTRACT FROM AUTHOR]

In the image analysis of the hydrophobic flocculation process of coal particles, the recognition accuracy of floc characteristic parameters is affected by bubbles, floc superposition, and floc morphology. An image processing method was studied and designed based on the PyCharm Community platform for clarity recognition and floc segmentation of coal particle floc characteristics. Firstly, the critical value of image clarity was determined to be 1.31 by using the Laplace convolution accounting method after the average gray level correction, and the fuzzy flocs whose clarity was less than the critical value were eliminated. Then, by comparing the segmentation effect of the threshold, edge, and region segmentation algorithm on the flocs, the histogram bimodal method was determined to be the best image processing segmentation method for fine coal floc collection. The fractal theory was used to verify the effect of image processing. The results showed that the surface fractal dimension of the floc image after deblurring and segmentation was maintained near 2.05, indicating that the processed image can obtain more stable and reliable floc characteristic parameters. [ABSTRACT FROM AUTHOR]

In recent years, artificial intelligence technology has seen increasingly widespread application in the field of intelligent manufacturing, particularly with deep learning offering novel methods for recognizing geometric shapes with specific features. In traditional CNC machining, computer-aided manufacturing (CAM) typically generates G-code for specific machine tools based on existing models. However, the tool paths for most CNC machines consist of a series of collinear motion commands (G01), which often result in discontinuities in the curvature of adjacent tool paths, leading to machining defects. To address these issues, this paper proposes a method for CNC system machining trajectory feature recognition and path optimization based on intelligent agents. This method employs intelligent agents to construct models and analyze the key geometric information in the G-code generated during CNC machining, and it uses the MCRL deep learning model incorporating linear attention mechanisms and multiple neural networks for recognition and classification. Path optimization is then carried out using mean filtering, Bézier curve fitting, and an improved novel adaptive coati optimization algorithm (NACOA) according to the degree of unsmoothness of the path. The effectiveness of the proposed method is validated through the optimization of process files for gear models, pentagram bosses, and maple leaf models. The research results indicate that the CNC system machining trajectory feature recognition and path optimization method based on intelligent agents can significantly enhance the smoothness of CNC machining paths and reduce machining defects, offering substantial application value. [ABSTRACT FROM AUTHOR]

Accurate determination of microscopic parameters is crucial for employing the discrete element method in addressing practical engineering challenges. The angle of repose calibration method for bulk materials is employed but frequently relies on subjective human measurements, potentially resulting in errors. This paper introduces a parameter calibration method that utilises a convolutional neural network to enhance standardisation, universality, and accuracy in predicting particle material behaviour. Firstly, the angle of repose simulations are conducted to establish training and test datasets. Next, sensitivity analysis is performed to determine the evaluation index. Subsequently, the performance differences in prediction accuracy among various input data types and network models, including one-dimensional convolutional, two-dimensional convolutional, and fully connected networks were compared. Finally, the influence of particle size and material type on the trained network model was investigated. The experimental results demonstrate that convolutional neural networks outperform traditional parameter calibration methods, in terms of feature extraction capabilities. According to the evaluation indicators in this paper, the conventional method achieves the highest prediction accuracy of 63.33%, whereas the deep learning method achieves a prediction accuracy of 86.67%. Additionally, the accuracy of one-dimensional convolutional network predictions is relatively high when compared to two-dimensional convolutional and fully connected networks. Furthermore, contour feature data exhibits superiority over slope data. Specifically, when the network input data consists of contour data, the prediction accuracy is further enhanced by 6.67% due to its inclusion of more effective features. This study provides new insights into the angle of repose parameter calibration. [Display omitted] • A parameter calibration method based on convolutional neural network. • This method does not require measuring the angle of repose. • This method is more standardised, unified, and accurate. • This method has a certain degree of interpretability. [ABSTRACT FROM AUTHOR]

The paper proposes a non-iterative training algorithm for a power efficient SNN classifier for applications in self-learning systems. The approach uses mechanisms of preprocessing of signals from sensory neurons typical of a thalamus in a diencephalon. The algorithm concept is based on a cusp catastrophe model and on training by routing. The algorithm guarantees a zero dispersion of connection weight values across the entire network, which is particularly important in the case of hardware implementation based on programmable logic devices. Due to non-iterative mechanisms inspired by training methods for associative memories, the approach makes it possible to estimate the capacity of the network and required hardware resources. The trained network shows resistance to the phenomenon of catastrophic forgetting. Low complexity of the algorithm makes in-situ hardware training possible without using power-hungry accelerators. The paper compares the complexities of hardware implementations of the algorithm with the classic STDP and conversion procedures. The basic application of the algorithm is an autonomous agent equipped with a vision system and based on a classic FPGA device.

Currently, the results of lung sound auscultation with either physical or electronic stethoscopes still rely mainly on the doctor's professional auscultation identification ability, which has not yet been able to realise intelligent diagnosis and interpretation. When patients are affected by lung diseases at home, they are unable to detect lung abnormalities on their own and delay treatment; when they are in the process of rescue and treatment of respiratory infectious diseases, in-ear stethoscopes are easily contaminated and cause nosocomial infections. Although stethoscopic sounds contain a wealth of information about health status, the lack of standardised collection methods, classification criteria and analysis tools has limited the objective analysis and application of stethoscopic sounds in practice. In this study, the data collection, arrangement and database design of the lung auscultation sound were carried out by using the unified auscultation sound collection equipment and process. The study used the software MetlabR2017a for data management and analysis to create a database of lung auscultation sounds in a healthy group and a group of patients with lung disease. A database of lung auscultation sounds was established for healthy groups and groups of patients with lung diseases. A standard set of classification of auscultatory tones, labelling specifications, audio characteristic signal parameters were developed. Building a system for storing, managing and analysing lung auscultation sound data to provide important data support for research related to the screening and monitoring of lung diseases and the translation of medical artificial intelligence applications. The study accumulated the experience of building an audio database of lung auscultation sounds, provided a useful reference for the management and analysis of the audio database, and laied the foundation for supporting the subsequent application of medical artificial intelligence-assisted auscultation in the screening and monitoring of lung diseases, which was of great medical value and practical application.

Aiming at the demand for accurate damage assessment of building targets, a new damage assessment method based on finite element simulation results recognition is proposed. The structural dynamic finite element analysis software SAP-2000 is used for numerical simulation and analysis of target damage, and the pre-assessment of target damage before attack is realized by numerical simulation images feature recognition and quantization combined with the target functional and physical damage level discrimination criteria. The rationality and availability of the method are verified by the simulation of examples.

Aiming at the demand for accurate damage assessment of building targets, a new damage assessment method based on finite element simulation results recognition is proposed. The structural dynamic finite element analysis software SAP-2000 is used for numerical simulation and analysis of target damage, and the pre-assessment of target damage before attack is realized by numerical simulation images feature recognition and quantization combined with the target functional and physical damage level discrimination criteria. The rationality and availability of the method are verified by the simulation of examples. [ABSTRACT FROM AUTHOR]

Copyright of e-Education Research is the property of Northwest Normal University and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In this paper, we construct an enterprise economic management sharing platform based on the feature recognition grid system, firstly, we use the fuzzy matching recognition algorithm to extract the grid features of economic management and map them into the fuzzy set of affiliation features, and then we extract the feature vectors from each grid to get the feature matrix, and finally, according to the similarity degree of the feature sequences, we set a threshold to give the recognition result. The results show that the recognition rate of the platform constructed in this paper reaches more than 99%, the classification accuracy is about 90% on average, and the optimization time is only about 19 s. It is verified that the method of this paper improves the accuracy of enterprise economic management, and has a positive role in promoting the healthy development of enterprises. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Chongqing University of Technology (Natural Science) is the property of Chongqing University of Technology and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Incorporating style-related objectives into shape design has been centrally important to maximize product appeal. However, algorithmic style capture and reuse have not fully benefited from automated data-driven methodologies due to the challenging nature of design describability. This paper proposes an AI-driven method to fully automate the discovery of brand-related features. First, to tackle the scarcity of vectorized product images, this research proposes two data acquisition workflows: parametric modeling from small curve-based datasets, and vectorization from large pixel-based datasets. Second, this study constructs BIGNet, a two-tier Brand Identification Graph Neural Network, to learn from both scalar vector graphics' curve-level and chunk-level parameters. In the first case study, BIGNet not only classifies phone brands but also captures brand-related features across multiple scales, such as lens' location, as confirmed by AI evaluation. In the second study, this paper showcases the generalizability of BIGNet learning from a vectorized car image dataset and validates the consistency and robustness of its predictions given four scenarios. The results match the difference commonly observed in luxury versus economy brands in the automobile market. Finally, this paper also visualizes the activation maps generated from a convolutional neural network and shows BIGNet's advantage of being a more explainable style-capturing agent. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Geodesy & Geodynamics (1671-5942) is the property of Editorial Board Journal of Geodesy & Geodynamics and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Feature recognition (FR) is one of the main tasks involved in computer-aided design, computer aided process planning, and computer-aided manufacturing systems. Conventional FR methods have topology, voxel, and pixel as model input data, which are rule-based, body decomposition-based, and neural network-based, respectively. However, FR methods are mostly applied to identify geometric features and are rarely manufacturing oriented. Recognizable feature types depend on the establishment of a feature database, which can easily lead to complex FR errors or omissions. This study proposes a novel recognition method for the general machining feature of 2.5-axis, one of the basic and commonly encountered feature types in manufacture industries. A novel ray fading algorithm is proposed to calculate the feature machining direction, and the type of 2.5-axis machining features is determined by both machining direction and topology. Features with machining directions can effectively assist the intelligent process planning to reduce the clamping changes and can potentially lead to significant time reduction for part machining. [ABSTRACT FROM AUTHOR]

As a cross-perception and cognitive research field in video understanding, motion training feature recognition is a very challenging task to establish a good spatio-temporal modeling of human motion due to the uncertainty of human motion speed, start and end time, appearance and posture, as well as the interference of physical factors such as lighting, perspective and occlusion. The purpose of this study is to use artificial intelligence data mining technology to study the feature recognition application of iot voice devices in sports training. Install the sensor in the appropriate position according to the position and posture to be measured. Ensure that the sensor can accurately measure the relevant features and maintain a stable connection. Using iot voice devices for data acquisition, sensors collect data on relevant features in real time to transmit the data to a cloud platform or local processing device via a wireless connection. By analyzing and mining the data collected by iot voice devices, we hope to effectively identify the characteristics of sports training and provide accurate feedback and guidance for athletes and coaches. The experimental results show that the iot voice device based on artificial intelligence data mining has achieved good results in the feature recognition application of sports training. Through the analysis of sports training data, we can successfully identify the characteristic patterns of different movements, and accurately predict the athletic state and posture of athletes.

The core extraction mechanism is one of the key components of a mold, and its design accuracy not only affects the production quality of molded parts, but also plays a key role in the service life of the mold. The traditional core extraction mechanism design process has low automation, tedious steps, difficult work, and can lead to a series of problems such as long mold design cycle and low production efficiency. Therefore, this study introduces intelligent CAD technology into the core extraction mechanism design process and develops a complete human-computer interaction design interface for the inclined column side core extraction mechanism, which is applicable to both die-casting and injection molds and greatly improves the efficiency of mold design. Firstly, we designed the corresponding feature recognition algorithm by two methods of "graph-based feature recognition" and "surface-based feature recognition", and developed the intelligent feature recognition subsystem. Secondly, based on the feature data of the molded parts and the principle of core extraction mechanism design, the parametric design module of each component of the lateral core extraction mechanism of the inclined guide pillar is developed. Finally, the reliability and rationality of the parametric drive design of the system are verified by the CAE motion simulation module of UG. The system is applicable to the design of core extraction mechanism of die-casting and injection molds, which greatly simplifies the tedious work of modeling the core extraction mechanism of inclined guide pillar and improves the efficiency of mold design.

In recent years, artificial intelligence technology has seen increasingly widespread application in the field of intelligent manufacturing, particularly with deep learning offering novel methods for recognizing geometric shapes with specific features. In traditional CNC machining, computer-aided manufacturing (CAM) typically generates G-code for specific machine tools based on existing models. However, the tool paths for most CNC machines consist of a series of collinear motion commands (G01), which often result in discontinuities in the curvature of adjacent tool paths, leading to machining defects. To address these issues, this paper proposes a method for CNC system machining trajectory feature recognition and path optimization based on intelligent agents. This method employs intelligent agents to construct models and analyze the key geometric information in the G-code generated during CNC machining, and it uses the MCRL deep learning model incorporating linear attention mechanisms and multiple neural networks for recognition and classification. Path optimization is then carried out using mean filtering, Bézier curve fitting, and an improved novel adaptive coati optimization algorithm (NACOA) according to the degree of unsmoothness of the path. The effectiveness of the proposed method is validated through the optimization of process files for gear models, pentagram bosses, and maple leaf models. The research results indicate that the CNC system machining trajectory feature recognition and path optimization method based on intelligent agents can significantly enhance the smoothness of CNC machining paths and reduce machining defects, offering substantial application value.

Introduction: Vision-based technologies are being widely used due to the availability of high-resolution cameras and open-source image processing tools. In the present work, a smartphone is used to capture video of a simple pendulum from various distances. Method: Image processing techniques have been used to analyse the motion. Subtle movements of the pendulum that were indistinct to the human eye (as observed from 10 m) were acquired as time–displacement data, and the Fast Fourier transform (FFT) was performed to determine the frequency of the moving pendulum. Results and Discussion: A comparative study for the two cases of pendulum motion, captured along the line of sight and perpendicular to the line of sight of the mobile camera has been presented. The observation distance of the mobile camera from the pendulum was varied in the range of 0.75–10 m. The results obtained were validated against the output from the laser vibrometer and analytical solution. The experiments demonstrate the smartphone camera's capability as a vibration measurement tool for a distant object. [ABSTRACT FROM AUTHOR]

With the active development of coal-related industries in Ningdong, problems such as land degradation caused by coal-based site expansion and soil pollution caused by coal-based solid waste discharge are getting progressively worse. The identification of space types and the proposal of risk management are the key basic issues of regional energy resource low-carbon development and ecological protection. This article proposes an index system for feature extraction at the regional and site scales. The spatial–temporal evolution trends and differences in Ningdong coal power base from 2003 to 2021 were identified and interpreted through feature recognition based on Landsat images at the regional scale. Accurate site type recognition was conducted based on 10 m resolution Sentinel-2 images from 2021. We utilized the PSR (pressure–state–response) model for a comprehensive assessment of risk management in Ningdong. The results of this study show that the coal site > the coal chemical site > the coal power site, and the risk of the coal power base is in a controllable state; thus, we put forward a zoning control strategy. Our data on the pollution risk management of large-scale coal-fired power generation complexes are of significant importance for site remediation and regional ecological restoration. [ABSTRACT FROM AUTHOR]

The development of China's solid wood furniture industry is relatively mature. As a precious raw wood material, rosewood plays an important role in the solid wood furniture industry. The number of imported rosewood in China has also increased year by year. The traditional identification method of rosewood categories mainly relied on the experience of professionals, so accurate and scientific identification of rosewood was very important for both furniture industry and handicrafts. In this article, a rosewood recognition method based on the surrounding features of wood was proposed. The BP neural network algorithm was used to establish a recognition model, which has been proven to have good recognition effect and can provide a new method for the identification of rosewood. [ABSTRACT FROM AUTHOR]

Numerous attempts have been made to recognize the machining features in three-dimensional (3D) computer-aided design (CAD) models using various methods since the 1980s. Recently, deep learning approaches have been explored for machining feature recognition. However, the boundary representation (BRep) model, the most common representation of 3D CAD models, is difficult to use directly in deep learning because of its complex structure. To solve this problem, BRepNet was recently proposed. This study proposes a method for recognizing machining features in 3D CAD models represented by BRep using BRepNet. In the proposed method, BRepNet is used to classify each face of a 3D CAD model based on the machining features. Next, the classified faces are combined into machining features using connected-component analysis. In addition, a dataset is generated to train the BRepNet model. Subsequently, the proposed method is implemented and tested for verification. The proposed method exhibits an accuracy of 96.03 % and part intersection over union (pIoU) of 90.57 %. [ABSTRACT FROM AUTHOR]

The utilization of moving image feature extraction in sports teaching has garnered increasing attention. However, traditional feature extraction algorithms often struggle to meet the diverse and complex demands of moving images. To address this challenge, this paper proposes a multi-scale feature fusion algorithm aimed at improving feature extraction in moving images. The algorithm begins by decomposing the moving image into multiple scales, followed by extracting features from each scale using a feature extraction network. To obtain a more comprehensive and accurate feature representation, feature fusion technology is employed to merge the features from different scales. The proposed algorithm, based on multi-scale feature fusion, exhibits a significant improvement in both accuracy and stability when compared to traditional feature extraction algorithms. Byaccurately extracting and representing the crucial features within moving images, the algorithm contributes to an improved understanding of athletes' movements, enabling instructors to provide more targeted and insightful feedback. This algorithm effectively captures key features within the moving images, providing robust support for tasks such as movement analysis and skill evaluation in sports teaching.

Planning of laser scanning path is the key process strategy of selective laser melting (SLM) additive manufacturing technology. Combining the significant structural characteristics of a three-dimensional model to realize process path planning is an important measure to improve the forming quality of parts. An SLM regional path planning method based on suspension recognition was proposed. By identifying the suspension characteristic region of the three-dimensional model, combined with the contour offset algorithm, the forming region segmentation and scanning path planning are realized. The effects of scanning line angle, interlayer rotation and different offset distance on the forming quality of suspension structure characteristics were studied by means of numerical simulation and process test. The results show that when the overhanging edge is offset by a reasonable distances and the scanning strategy parallel to the overhanging edge is adopted in this area, the deformation and residual stress on the overhanging edge can be reduced by 54% and 73% at most.

With the construction of new-type power system under the ”double carbon” target and the increasing diversification of the energy demand of the user side, the short-term load forecasting of power system is facing new challenges. In order to fully exploit the massive information contained in big data, this paper proposed a new short-term load forecasting method for new-type power system considering multiple factors, which based on Graph Convolutional Network (GCN) and Long Short-Term Memory network (LSTM). Spearman rank correlation coefficient was used to analyze the correlation between load and meteorological factors, and a quantitative model including meteorological factors, date factors and regional factors was established. Thus, GCN and LSTM were jointly used to extract the spatial and temporal characteristics of massive data respectively, and finally the short-term power load forecasting was achieved. The public data sets were used for performance verification compared with three comparison models, LSTM, CNN-LSTM and TCN-LSTM. The results show that the proposed method can make full use of the influence of multi-dimensional data, meanwhile improve the load prediction accuracy and training efficiency effectively.

Due to the increasingly complex environment of urban gas pipeline service，community gas pipeline leakage events occur frequently，so，effective identification of hot spots for corrosion in community gas pipelines is crucial for the safe operation of community gas pipelines.In this paper，field tests and experiments were carried out in communities A，B and C，and service parameters such as test piece power-on/off potential，test piece current density，ground potential gradient，and soil resistivity of pipelines in communities A，B and C were tested，and the corrosion products of the polarization inspection piece connected to the community A pipeline were analyzed by XRD.Besides，the electrical connectivity between the gas pipeline and the surrounding steel structures in community A was tested.Results showed that the gas pipelines in communities A，B and C all had a significant positive pipe-to-soil potential，the power-on pipe-to-soil potential was significantly positive compared to the power-off potential，and there was a steady current outflow.The corrosion products of the polarization inspection piece of the pipeline in community A were mainly Fe3O4，SiO2and Fe2MgO4，in which SiO2mainly came from the soil on the surface of the polarization inspection piece.In addition，there existed the electrical connection between the gas pipeline in community A and the surrounding steel structures，which accelerated the corrosion of the pipeline.

With the increasing importance of music information retrieval, the construction of effective music recognition methods has gradually become one of the research focuses. The vocal map features of the collected music pronunciation signals are extracted. The research in this paper is primarily based on the basic physical characteristics of music notes; once these characteristics are identified, then they are mathematically extracted and analysed (voiceprint feature method), and a recognition model is established. On this basis, the process of adaptive separation and recognition of music signals is completed. Finally, the performance of different recognition types is verified and evaluated through experiments. The results show that the average accuracy rate of the modified algorithm within a certain range reaches 73.6%; additionally, the average accuracy rate is increased by 10.95% compared with the audio recognition based on Internet of Things (IoT) data, and is more accurate than the audio recognition method based on data collection, showing an improvement of 20.75%. This shows that the modified recognition algorithm adopted in the present research has stable and high accuracy for comprehensive music types with different characteristics. Finally, the identification method proposed in this paper shortens the time by 85.71% compared with the identification method of data collection, and shortens the identification time by 83.33% compared with the IoT identification method. This greatly improves the recognition of different musical feature types.

This study investigates the application of virtual reality technology in the creative expression of digital film and television productions, especially the role of EEG signal denoising and feature recognition methods in enhancing the audience experience. The study uses wavelet threshold denoising and parallel RLS adaptive filtering algorithms to process EEG signals to improve the accuracy and reliability of the data. Then, the EEG signals were feature extracted using a bihemispheric domain adversarial neural network (BiDANN) to more accurately recognize the user’s emotional responses. The experimental results show that in the virtual reality environment, the users’ concentration and emotional reactions are significantly improved, with the average concentration reaching 74.21 and the average value of the electrodermal test data being 6.19. In addition, the eye-movement interaction experiments show that different types of digital movie and television works can cause additional attention allocation of users in the VR environment, leading other creative performance effects. The study’s results prove that virtual reality technology can significantly enhance the innovative performance of digital movie and television works and improve the audience’s viewing experience.

Architecture is a representative of a city. It is also a spatial carrier of urban culture. Identifying the architectural features in a city can help with urban transformation and promote urban development. The use of visual saliency models in regional architectural texture recognition can effectively enhance the effectiveness of regional architectural texture recognition. In this paper, the improved visual saliency model first enhances the texture images of regional buildings through histogram enhancement technology, and uses visual saliency algorithms to extract the visual saliency of the texture features of regional buildings. Then, combined with the maximum interclass difference method of threshold segmentation, the visual saliency image is segmented to achieve accurate target recognition. Finally, the feature factor iteration of the Bag of Visual Words model and the function classification of support vector machines were used to complete the recognition of regional architectural texture features. Through experimental verification, the constructed regional architectural texture feature recognition method based on visual saliency model can effectively enhance the recognition image. This method performs well in boundary contour separation and visual saliency, with an average recognition rate of 0.814 for texture features in different building scenes, indicating high stability. [ABSTRACT FROM AUTHOR]

Sports news is a type of discourse that is characterized by a specific vocabulary, style, and tone, and it is typically focused on conveying information about sporting events, athletes, and teams. Thematic context-based deep learning is a powerful approach that can be used to analyze and interpret various forms of natural language, including the discourse expression of sports news. An application model of sign language and lip language recognition based on deep learning is proposed to facilitate people with hearing impairment to easily obtain sports news content. First, the lip language recognition system is constructed; next, MobileNet lightweight network combined with Long-Short Term Memory (LSTM) is used to extract lip reading features. ResNet-50 residual network structure isadopted to extract the features of sign language; finally, the convergence, accuracy, precision and recall of the model are verified respectively. The results show that the loss of training set and test set converges gradually with the increase of iteration times; the lip language recognition model and the sign language recognition model basically tend to be stable after 14 iterations and 12 iterations, respectively, suggesting a better convergence effect of sign language recognition. The accuracy of sign language recognition and lip language recognition is 98.9% and 87.7%, respectively. In sign language recognition, the recognition accuracy of numbers 1, 2, 4, 6 and 8 can reach 100%. In lip language recognition, the recognition accuracy of numbers 2, 3 and 9 is relatively higher. This exploration can facilitate hearing-impaired people to quickly obtain the relevant content in sports news videos, and also provide help for their communication. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Materials Engineering / Cailiao Gongcheng is the property of Journal of Materials Engineering Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of China Mechanical Engineering is the property of Editorial Board of China Mechanical Engineering and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

With the development of information technology, computer technology is also gradually applied to the teaching activities of art education, and the use of multimedia technology to assist music teaching has become one of the hot research areas in universities. In order to better cultivate university students' musical exploration ability and creativity, a music note feature recognition teaching model based on Hidden Markov Model (HMM) algorithm is studied and optimized in universities by using genetic algorithm based on HMM algorithm. The music note feature recognition teaching model studied in this article combines computer multimedia technology, signal processing technology, and music theory, and uses computers to simulate the process of human cognition and analysis of music. And in this article, a music note recognition system was constructed using the features of sound level contours combined with the HMM algorithm. The data extracted during music recognition was compressed using the energy compression feature of sound level contours. At the same time, maximum likelihood estimation was used to find the optimal chord sequence, i.e., the optimal path, for the input signal. In the experimental results, the minimum value of the objective function was about 0.739 when the variance probability and crossover probability were 0.02 and 0.6, respectively. In the results, the HMM algorithm-based music note recognition model can improve the quality of music teaching and has some potential for application in the field of music teaching.

Presently, for the identification and classification of images, various deep learning techniques are being used. In these techniques, the whole image is considered to produce similar feature sets for many images. As a result, this mechanism loses many of its features at the final stage. Therefore, to analyze and identify medicinal leaves through an artificial eye of botanists, it was emphasized that the leaf image features should remain preserved till the final stage of classification for better accuracy. The existing plant identification approaches are trained using the leaf images. So leaf features are lost in the different stages of the convolution process and the same feature values are generated for similar type leaf images. This raises ambiguity in the results and affects the accuracy of leaf image identification. But here, in this proposed deep learning-based plant leaves morphological feature recognition system, leaf morphological features are used to train the system. Morphological features are identified to recognize a plant leaf. Here, morphological features of medicinal plant leaves, venation, shapes, apices, and bases are extracted and analyzed to predict the image class. So, the leaf features remain persevered until the final stage. The proposed feature recognition analysis improves the accuracy of the leaf identification method. In this, more than 300 leaves from 18 different plant families are collected and trained to build the deep learning classifier and achieve 96% accuracy. The performance evaluation was also conducted over "Flavia", "Swedish" and "Leaf" data set and obtained 91%, 87% and 91% accuracy. The performance of image classification and feature preservation algorithms with less computational power are indicating the potential applicability of the proposed Deep - Morpho Algorithm (DMA) in medicinal plants and leaves identification. [ABSTRACT FROM AUTHOR]

In order to solve the problem of autonomous movement of disabled people brain computer interface (BCI) has been applied in the exoskeleton widely. In the practical use, the low signal-noise ratio of electroencephalogram (EEG) signal results in the low classification accuracy in BCI. In order to improve the signal recognition rate of lower limb exoskeleton based on brain computer interface, particle swarm optimization support vector machine (PSO-SVM) algorithm was used to improve the EEG signal recognition rate, and 86. 52% EEG signal recognition rate was achieved. Firstly, the common spatial pattern (CSP) mathematical model was established for feature extraction of EEG signals, and then a particle PSO-SVM classification model was established. Secondly, the key parameters of EEG classification were optimized, and the final experimental data were compared with the traditional SVM classification method. Finally, the algorithm was verified and the lower limb exoskeleton experiment was carried out. The experimental results show that the classification accuracy of PSO-SVM is significantly higher than that of traditional SVM, and the average classification result can reach 86. 52%, which improves the recognition rate of motor imagery (MI) EEG signals. The proposed method of feature recognition of MI signals based on PSO-SVM, which can realize the accurate recognition of MI, and provide theoretical basis and technical support for the application of brain computer interface technology in the field of exoskeleton. [ABSTRACT FROM AUTHOR]

In the context of additive manufacturing, the adjustment of process data to individual geometric features offers the potential to further increase manufacturing speed and quality, while being widely underestimated in recent research. Unfortunately, the current non-uniformd at a handling in the CAD-CAM-Link results in adown stream data loss, that prevents the availability of geometric knowledge from being present at any time to apply the more advanced approaches of adaptive slicing and tool path generation. Automatic detection of various geometric entities would be beneficial for classifying partial surfaces and volumetric ranges to gain customized informational insights of geometric parameterization. In this work, an enhanced approach of geometric deep learning for the analysis of voxelized engineering parts will be presented to align the inference representations to modeling paradigms for complex design models like architected materials. Although the baseline voxel representation offers distinct advantages in detection accuracy, it comes with an adversely large memory footprint. The geometry discretization leads to high resolutions needed to capture various detail levels that prevent the analysis of fine-grained objects. To achieve efficient usage of three-dimensional (3D) deep learning techniques, we propose a 3D-convolutional neural network-based feature recognition approach using signed distance field data to limit the needed resolution. These implicit geometric data leverage the advantages of volumetric convolution while alleviating their disadvantages through the use of the continuous signed distance function. When analyzing computer-aided design data for geometric primitive features, a common application task in surface reconstruction of reverse engineering the proposed methodology, achieves a detection accuracy that is in line with the accuracy values achieved by comparable algorithms. This enables the recognition offine-grained surface instances. The unambiguous shape information extracted could be used in subsequent adaptive slicing algorithms to achieve individual geometry-based hatch generation. [ABSTRACT FROM AUTHOR]

The automatic P-wave onset time (P-onset) picking of microseismic (MS) waveforms generated during rock failure is the basis of and key to locating the source and exploring the failure mechanism of rock failure in underground engineering. However, it is hard to ensure the accuracy in picking P-onset of MS waveforms with a low signal-to-noise ratio (SNR) mixed with noises. Moreover, traditional P-onset picking methods mainly depend on manual intervention, which imposes onerous requirements on the SNR of MS waveforms. Hence, a new P-onset picking method is proposed based on the deep learning model: the principal components of MS waveforms are extracted based on the multi-channel singular spectrum analysis (MSSA) method to reduce the influence of noise; afterwards, a short-time Fourier transform (STFT) is applied to the processed MS waveforms to attain essential parameters of the waveforms; finally, a model for P-onset picking is established based on the long short-term memory (LSTM) network. Subsequently, a comparison is conducted between the proposed method and other methods based on actual field MS data collected from Xiadian Gold Mine and the seismic data from Stanford Earthquake Dataset. The results show that the proposed method can accurately extract data features of MS waveforms and further improve the P-onset picking performance.