Recognition Algorithm Research Articles

The Development of Spinal Endoscopic Ultrasonic Imaging System with an Automated Tissue Recognition Algorithm.

Preclinical experimental study. To develop an intraoperative ultrasound-assisted imaging device, which could be placed at the surgical site through an endoscopic working channel and which could help surgeon recognition of different tissue types during endoscopic spinal surgery (ESS). ESS remains a challenging task for spinal surgeons. Great proficiency and experience are needed to perform procedures such as intervertebral discectomy and neural decompression within a narrow channel. The limited surgical view poses a risk of damaging important structures, such as nerve roots. We constructed a spinal endoscopic ultrasound system, using a 4-mm custom ultrasound probe, which can be easily inserted through the ESS working channel, allowing up to 10 mm depth detection. This system was applied to ovine lumbar spine samples to obtain ultrasound images. Subsequently, we proposed a two-stage classification algorithm, based on a pretrained DenseNet architecture for automated tissue recognition. The recognition algorithm was evaluated using accuracy and consistency. The probe can be easily used in the ESS working channel and produce clear and characteristic ultrasound images. We collected 367 images for training and testing of the recognition algorithm, including images of the spinal cord, nucleus pulposus, adipose tissue, bone, annulus fibrosus and nerve roots. The algorithm achieved over 90% accuracy in recognizing all types of tissues with a Kappa value of 0.875. The recognition times were under 0.1 s using the current configuration. Our system was able to be used in existing ESS working channels and clearly identified at-risk spinal structures in vitro. The pretrained algorithms could identify six intraspinal tissue types accurately and quickly. The concept and innovative application of intraoperative ultrasound in ESS may shorten the learning curve of ESS and improve surgical efficiency and safety.

Real time detection and identification of fish quality using low-power multimodal artificial olfaction system

Single gas quantification and mixed gas identification have been the major challenges in the field of gas detection. To address the shortcomings of chemo-resistive gas sensors, sensor arrays have been the subject of recent research. In this work, the research focused on both optimization of gas-sensing materials and further analysis of pattern recognition algorithms. Four bimetallic oxide-based gas sensors capable of operating at room temperature were first developed by introducing different modulating techniques on the sensing layer, including constructing surface oxygen defects, polymerizing conducting polymers, modifying Nano-metal, and compositing flexible substrates. The signals derived from the gas sensor array were then processed to eliminate noise and reduce dimension with the feature engineering. The gases of were qualitatively identified by support vector machine (SVM) model with an accuracy of 98.86 %. Meanwhile, a combined model of convolutional neural network and long short-term memory network (CNN-LSTM) was established to remove the interference samples and quantitatively estimate the concentration of the target gases. The combined model based on deep learning, which avoids the overfitting with local optimal solutions, effectively boosts the performance of concentration recognition with the lowest root mean square error (RMSE) of 2.3. Finally, a low-power artificial olfactory system was established by merging the multi-sensor data and applied for real-time and accurate judgment of the food freshness.

Identifying Missing Individual using AI

In recent days, the daily increase in many of the missing persons has made it increasingly challenging to locate them. To get over this challenge, machine learning algorithms, particularly facial recognition, can be employed to identify missing individuals. This approach aims to simplify the search process for both parents or guardians and the police. In this system, parents or guardians of the missing person upload a photo, which is stored in a database. The facial recognition algorithm then utilizes k-nearest neighbors (k-NN) to search for a match within the database. If a match is detected, both the police and the parents or guardians are alerted. Experimental outcomes tells that users appreciate the new features of the application and find the system user-friendly

Open-Vocabulary Part-Level Detection and Segmentation for Human–Robot Interaction

Object detection and segmentation have made great progress in robotic application missions. However, intelligent agents require fine-grained recognition algorithms rather than object-level and language instructions to enhance the interaction between humans and robots. To improve the robot’s interactivity in the process of the robot response to language instructions, we propose a method for part-level detection and segmentation by exploiting vision language models. In this approach, Swin Transformer is introduced in the image encoder for extracting image features, and FPNs (Feature Pyramid Networks) are modified to better process the features from Swin Transformer. Next, the image decoder is proposed for model aligning between the image features and text embeddings for achieving human–robot interaction via language. Finally, we verify that the text embeddings are impacted by the command of input and that different prompt templates also affect classification. The method proposed in this paper, which is validated on two datasets (PartImagePart and Pascal Part), possesses the ability to understand and execute part-level missions and accurately segments and detects parts compared with existing interactive methods.

Real-time object detection, tracking, and monitoring framework for security surveillance systems

The concept of security is becoming a global challenge, and governments, stakeholders, corporate societies, and individuals must urgently create a reasonable protection mechanism for good. Therefore, a real-time surveillance system is essential for detection, tracking, and monitoring. Many studies have attempted to provide better solutions but more research and better approaches are essential. This study presents a real-time framework for object detection and tracking for security surveillance systems. The system has been designed based on approximate median filtering, component labeling, background subtraction, and deep learning approaches. The new algorithms for object detection, tracking, and recognition have been implemented using Python and integrated with C# programming languages for ease of use. A software application framework is designed, implemented, and evaluated. The experimental results based on MOT-Challenge performance metrics show that the proposed algorithms have much better performance in terms of accuracy and precision on the MOT15, MOT16, and MOT17 datasets compared to state-of-the-art approaches. This framework also provides an accurate and effective means of monitoring and recognizing moving objects. The software development, including the design of the framework user interfaces, is coded in the C# programming language and integrated with Python using Microsoft Visual Studio (2019 edition). The integration is performed to provide a convenient user interface and to enable the execution of the framework as a standard and standalone software application. Future studies will consider the dynamic scalability of the framework to accommodate different surveillance application areas in overcrowded scenarios. Multiple data sources are integrated to enhance the performance for different scene times, locations, and weather conditions. Furthermore, other object-detection techniques such as You Only Look Once (YOLO) and its variants shall be considered in future studies. These techniques allow the framework to adapt to complex situations in which security surveillance is challenging.

A Comprehensive Analysis of Algorithms in Multiple-Face Recognition

A Comprehensive Analysis of Algorithms in Multiple-Face Recognition

Immersive artificial intelligence technology based on entertainment game experience in simulation of psychological health testing for university students

At present, the mental health problems of college students are becoming more and more prominent, but the existing mental health detection methods have certain limitations. Therefore, the combination of entertainment and game experience with artificial intelligence technology in this paper can provide a more comprehensive and accurate mental health detection scheme. The research designed an immersive entertainment game experience platform, in which artificial intelligence technology was used to detect mental health. The platform is based on emotion recognition algorithms and artificial intelligence interaction technology to assess the level of mental health of users by analyzing their behavior and emotional expression in the game. The results show that immersive artificial intelligence technology based on entertainment and game experience can effectively simulate the mental health state of college students. Compared with traditional mental health assessment methods, this technique has obvious advantages in terms of accuracy and comprehensiveness. Users also showed a high degree of acceptance and participation for this entertaining mental health detection method. This technology can provide students with a more relaxed and interesting mental health assessment experience, and can also provide accurate and comprehensive assessment results, and provide scientific and effective guidance for mental health management and intervention of college students.

Subtle signals: Video-based detection of infant non-nutritive sucking as a neurodevelopmental cue

Non-nutritive sucking (NNS), which refers to the act of sucking on a pacifier, finger, or similar object without nutrient intake, plays a crucial role in assessing healthy early development. In the case of preterm infants, NNS behavior is a key component in determining their readiness for feeding. In older infants, the characteristics of NNS behavior offer valuable insights into neural and motor development. Additionally, NNS activity has been proposed as a potential safeguard against sudden infant death syndrome (SIDS). However, the clinical application of NNS assessment is currently hindered by labor-intensive and subjective finger-in-mouth evaluations. Consequently, researchers often resort to expensive pressure transducers for objective NNS signal measurement. To enhance the accessibility and reliability of NNS signal monitoring for both clinicians and researchers, we introduce a vision-based algorithm designed for non-contact detection of NNS activity using baby monitor footage in natural settings. Our approach involves a comprehensive exploration of optical flow and temporal convolutional networks, enabling the detection and amplification of subtle infant-sucking signals. We successfully classify short video clips of uniform length into NNS and non-NNS periods. Furthermore, we investigate manual and learning-based techniques to piece together local classification results, facilitating the segmentation of longer mixed-activity videos into NNS and non-NNS segments of varying duration. Our research introduces two novel datasets of annotated infant videos, including one sourced from our clinical study featuring 18 infant subjects and 183 h of overnight baby monitor footage. Additionally, we incorporate a second, shorter dataset obtained from publicly available YouTube videos. Our NNS action recognition algorithm achieves an impressive 95.8% accuracy in binary classification, based on 960 2.5-s balanced NNS versus non-NNS clips from our clinical dataset. We also present results for a subset of clips featuring challenging video conditions. Moreover, our NNS action segmentation algorithm achieves an average precision of 93.5% and an average recall of 92.9% across 30 heterogeneous 60-s clips from our clinical dataset.

A standalone and portable imaging detection system with embedded computing for automated defect inspection of microfluidic devices

The existing methods for defect detection in PDMS microfluidic chips typically involve complex image recognition algorithms or manual inspection and still lack efficiency and reliability. Although some automatic defect detection methods have been proposed in recent years, most of them still rely on external computation systems to deploy. To address these challenges, we propose an independent portable defect detection system with embedded computing for microfluidic devices. This portable system is completely self-contained, integrating an image acquisition module, a control panel module, a power module, and an embedded computing control module to realize chip detection, processing, and result display functions. Experimental results show that the system can effectively detect most of the commonly seen defects in PDMS-based microfluidic chips, proving to be more efficient and reliable than manual inspection. With the control of the embedded system, two detection methods: template matching (based on comparison with standard samples) and automatic defect detection (based on surface defect recognition) were used to identify defects in PDMS-based microfluidic chips. The proposed system can automatically inspect and analyze chips without the need for external laboratory support and can provide a promising solution for future microfluidic chip manufacturing and operation.

CT-YoloTrad: fast and accurate recognition of point-distributed coded targets for UAV images incorporating CT-YOLOv7

Artificial point-distributed coded targets owns unique coded sequence numbers that can be recognized automatically. To address the issue of decreasing recognition accuracy and efficiency of existing recognition methods in complicated circumstances, an improved object detection model for coded target acquisition from unmanned aerial vehicle (UAV) images, CT-YOLOv7, is proposed. This improved model is based on the original YOLOv7 model, replacing several Conv with partial convolution (PConv), while introducing the bi-level routing attention mechanism, and designing the CBS-R structure and CBS-PR structure. In addition, the loss function is replaced with WIOU loss function to further improve the model’s performance. Based on the above, the new recognition method of point-distributed coded targets for UAV images is organized as follows. Firstly, CT-YOLOv7 is embedded into the front-end of the classical coded targets recognition process (that is, the coded targets are first extracted). Then, the extraction results are fed into the classical recognition algorithm for recognition. Lastly, the recognition results are inverse-calculated back to the original image. The new method aims to focus the processing on the region of interest to achieve fast and accurate coded targets recognition for UAV images. The experimental results show that CT-YOLOv7’s detection accuracy is 90.83%, which improves the accuracy by 8.46% and reduces the computation by 11.54% compared to the original YOLOv7. By incorporating the CT-YOLOv7 model, the time consumption for coded target recognition of a single UAV image is 150–350ms, which improves the average efficiency by 3–5 times compared with the classical method. Furthermore, the proposed method can correctly recognize regions with shadows and noise, and the recognition accuracy is improved by 15%–40%. With the method proposed in this paper, the coded targets are expected to be applied into UAV photogrammetry or remote sensing to realize accurate and quasi-real-time recognition.

Predicting oil prices: A comparative analysis of machine learning and image recognition algorithms for trend prediction

This paper investigates the effectiveness of machine learning algorithms, including logistic regression, artificial neural networks, support vector machines, gradient boosting algorithms (XGBoost, ExtraTrees), random forests, and convolutional neural network (CNN) for trend prediction of daily spot oil prices across horizons of 1 to 8 days. We utilize a comprehensive set of features, including technical indicators, financial data, and volatility measures, to predict trends in closing prices. Our results reveal that the CNN model significantly outperforms other algorithms. This superior performance likely stems from CNN’s ability to capture visual patterns in price movements, potentially mimicking how traders identify trends.

Road surface crack detection and classification based on YOLOv5

The burgeoning need for sustainable urban transportation has magnified the importance of effective road infrastructure maintenance, particularly road crack detection. This study addresses the challenge of accurately detecting road surface cracks through automated methods. To accurately identify road surface cracks, we used the YOLOv5 image recognition algorithm, enhanced by comprehensive data augmentation and meticulous training strategies. The YOLOv5 proved superior to traditional methods in accuracy and recall while maintaining high computational speed and a smaller model size in experimental outcomes. The findings affirm that YOLOv5-based detection technologies significantly improve maintenance efficiency and traffic safety, with future work poised to further refine these approaches for greater robustness and integration into intelligent transportation systems.

Improved Facial Expression Recognition Algorithm Based on Local Feature Enhancement and Global Information Association

Facial expression recognition is the key area of research in computer vision, enabling intelligent devices to understand human emotions and intentions. However, recognition of facial expressions in natural scenes presents challenges due to environmental factors like occlusion and pose variations. To address this, we propose a novel approach that combines local feature enhancement and global information correlation. This method allows the model to learn both local and global facial features along with contextual information. By enhancing salient local features and exploring multi-scale facial expression features, our model effectively mitigates the impact of occlusion and pose variations, improving recognition accuracy. Experimental results demonstrate that our adapted model outperforms alternative algorithms in recognizing facial expressions under challenging environments, achieving recognition accuracies of 85.07% and 99.35% on the RAF-DB and CK+ datasets, respectively.

High-precision regression prediction of HCHO concentration based on gas sensors and FPNet

The leakage of HCHO gas has direct and irreversible damage on both human health and the environment. However, the current commercial electronic nose (E-nose) has low accuracy in predicting the concentration of HCHO. In this work, a sensor array consisting of 6 gas sensors was fabricated and combined with advanced deep learning algorithms to achieve accurate prediction of HCHO concentration. High-performance SnO2 sensing material with high specific surface area was synthesized by applying the Metal-Organic Frameworks (MOFs) sacrificial template, and the response of the SnO2 gas sensor was as high as 15.098, which is much higher than that of commonly used commercial sensors. In addition, the sensing performance of the SnO2 sensor, including response, selectivity and repeatability etc., were investigated. Afterward, a Formaldehyde Prediction Network (FPNet) model was proposed for processing the HCHO gas dataset and cosine annealing was added to gradually reduce the learning rate during training to help the model converge to the optimal solution. Furthermore, the residual structures with different number of layers were compared, and the 3-layer residual structure expressed the most accurate results of concentration prediction of the HCHO gas, with an MSE of 9.003, MAPE of 0.015, and R2 of 0.998. This study provides a new idea from sensor preparation to pattern recognition algorithm research, effectively solving the problems of the low response of commercial sensors to HCHO and the low accuracy of pattern recognition algorithms for E-nose datasets.

Breaks in the Arctic ice cover: from observations to predictions

Breaks (ruptures) and cracks is the distinguishing feature of any ice cover in the Arctic seas during the cold season and in the whole Arctic Basin throughout a year. The formation of them is a consequence of macro-deformation of the ice thickness. Investigating of the ice breaking in the Arctic begins with single visual observations during the ice aerial surveys in the 1940s and continues till nowadays using regular information from artificial Earth satellites. Processing of big volumes of satellite data and creating climatological datasets on breaks became possible owing to the development of algorithms for automatic identification of the ice breaks in images. Interpretation of the satellite images is based on the fundamental difference between physical properties of breaks and the surrounding consolidated ice. Algorithms for automatic recognition of ruptures using satellite data obtained in different wavelength ranges, including the use of artificial intelligence, are currently being developed. The main characteristics of breaks which are usually analyzed are as follows: the summarized area of them and its ratio to the total area of the ice field, the mean and maximum widths as well as the total length. The temporal and spatial variability of these characteristics is also considered. Such information is needed for solving problems of improving models of ice cover dynamics and modeling the interaction between the ocean and the atmosphere at high latitudes. A specific feature of publications of the Russian authors on this topic is the practical use of the results obtained for hydrometeorological support of navigation in ice. For the navigation purposes, the dominant orientation of the ruptures on the way of ships is of greatest importance. Operational and prognostic information about the orientation and extent of ruptures, including distribution of them in an ice field are the key data for choosing the optimal sailing route in the Arctic.

Abnormal Behavior Recognition Based on 3D Dense Connections.

Abnormal behavior recognition is an important technology used to detect and identify activities or events that deviate from normal behavior patterns. It has wide applications in various fields such as network security, financial fraud detection, and video surveillance. In recent years, Deep Convolution Networks (ConvNets) have been widely applied in abnormal behavior recognition algorithms and have achieved significant results. However, existing abnormal behavior detection algorithms mainly focus on improving the accuracy of the algorithms and have not explored the real-time nature of abnormal behavior recognition. This is crucial to quickly identify abnormal behavior in public places and improve urban public safety. Therefore, this paper proposes an abnormal behavior recognition algorithm based on three-dimensional (3D) dense connections. The proposed algorithm uses a multi-instance learning strategy to classify various types of abnormal behaviors, and employs dense connection modules and soft-threshold attention mechanisms to reduce the model's parameter count and enhance network computational efficiency. Finally, redundant information in the sequence is reduced by attention allocation to mitigate its negative impact on recognition results. Experimental verification shows that our method achieves a recognition accuracy of 95.61% on the UCF-crime dataset. Comparative experiments demonstrate that our model has strong performance in terms of recognition accuracy and speed.

Deep Visual Computing of Behavioral Characteristics in Complex Scenarios and Embedded Object Recognition Applications.

By leveraging artificial intelligence and big data to analyze and assess classroom conditions, we can significantly enhance teaching quality. Nevertheless, numerous existing studies primarily concentrate on evaluating classroom conditions for student groups, often neglecting the need for personalized instructional support for individual students. To address this gap and provide a more focused analysis of individual students in the classroom environment, we implemented an embedded application design using face recognition technology and target detection algorithms. The Insightface face recognition algorithm was employed to identify students by constructing a classroom face dataset and training it; simultaneously, classroom behavioral data were collected and trained, utilizing the YOLOv5 algorithm to detect students' body regions and correlate them with their facial regions to identify students accurately. Subsequently, these modeling algorithms were deployed onto an embedded device, the Atlas 200 DK, for application development, enabling the recording of both overall classroom conditions and individual student behaviors. Test results show that the detection precision for various types of behaviors is above 0.67. The average false detection rate for face recognition is 41.5%. The developed embedded application can reliably detect student behavior in a classroom setting, identify students, and capture image sequences of body regions associated with negative behavior for better management. These data empower teachers to gain a deeper understanding of their students, which is crucial for enhancing teaching quality and addressing the individual needs of students.

Line spectrum target recognition algorithm based on time‐delay autoencoder

AbstractEffective extraction of target features has always been a key issue in target recognition technology in the field of signal processing. Traditional deep learning algorithms often require extensive data for pre‐training models to ensure the accuracy of feature extraction. Moreover, it is challenging to completely remove noise due to the complexity of the underwater environment. A Time‐Delay Autoencoder (TDAE) is employed to extract ship‐radiated noise characteristics by leveraging the strong coherent properties of line spectrum. This approach eliminates the need for previous data to adaptively develop a nonlinear model for line spectrum extraction. The test data was processed using three distinct approaches, and plots of recognition accuracy curves at various signal‐to‐noise ratios were made. On the dataset utilised in the research, experimental results show that the proposed approach achieves over 75% recognition accuracy, even at a signal‐to‐noise ratio of −15 dB.

The Research of Multi-Node Collaborative Compound Jamming Recognition Algorithm Based on Model-Agnostic Meta-Learning and Time-Frequency Analysis

Deep learning has presented its spectacular potential in the jamming recognition field. Yet, sufficient samples required by normal deep learning analysis methods are not always available, especially in the 6G communication field. This situation appears to be more challenging in the communication field. In this article, Model-Agnostic Meta-Learning (MAML) is imported into the jamming recognition field in order to accomplish compound jamming recognition in the circumstances of few-shot learning. Further, the existing research on jamming recognition techniques is mostly based on single-node recognition. This technique cannot make full and efficient use of the jamming information collected. Therefore, this article adds a multi-node collaborative technique into the compound jamming recognition algorithm that is based on MAML and time-frequency analysis. Based on the fact that each cognitive node can recognize independently, the recognition results are be sent to the fusion center. The fusion center completes the fusion of the recognition results according to the majority rule. The experiments demonstrate that, with the fusion of the multi-node collaborative technique, the precision of compound jamming recognition in the condition of few-shot learning has been effectively improved.

A High-Performance Anti-Noise Algorithm for Arrhythmia Recognition.

In recent years, the incidence of cardiac arrhythmias has been on the rise because of changes in lifestyle and the aging population. Electrocardiograms (ECGs) are widely used for the automated diagnosis of cardiac arrhythmias. However, existing models possess poor noise robustness and complex structures, limiting their effectiveness. To solve these problems, this paper proposes an arrhythmia recognition system with excellent anti-noise performance: a convolutionally optimized broad learning system (COBLS). In the proposed COBLS method, the signal is convolved with blind source separation using a signal analysis method based on high-order-statistic independent component analysis (ICA). The constructed feature matrix is further feature-extracted and dimensionally reduced using principal component analysis (PCA), which reveals the essence of the signal. The linear feature correlation between the data can be effectively reduced, and redundant attributes can be eliminated to obtain a low-dimensional feature matrix that retains the essential features of the classification model. Then, arrhythmia recognition is realized by combining this matrix with the broad learning system (BLS). Subsequently, the model was evaluated using the MIT-BIH arrhythmia database and the MIT-BIH noise stress test database. The outcomes of the experiments demonstrate exceptional performance, with impressive achievements in terms of the overall accuracy, overall precision, overall sensitivity, and overall F1-score. Specifically, the results indicate outstanding performance, with figures reaching 99.11% for the overall accuracy, 96.95% for the overall precision, 89.71% for the overall sensitivity, and 93.01% for the overall F1-score across all four classification experiments. The model proposed in this paper shows excellent performance, with 24 dB, 18 dB, and 12 dB signal-to-noise ratios.