Recognition Speed Research Articles

Research on electronic nose for compound malodor recognition combined with artificial neural network and linear discriminant analysis

The foul odor of foul gas has many harmful effects on the environment and human health. In order to accurately assess this impact, it is necessary to identify specific malodorous components and levels. In order to meet the qualitative and quantitative identification of the components of malodorous gas, an electronic nose system is developed in this paper. Both principal component analysis (PCA) and linear discriminant analysis (LDA) were used to reduce the dimensionality of the collected data. The reduced-dimensional data are combined with a support vector machine (SVM) and backpropagation (BP) neural network for classification and recognition to compare the recognition results. Regarding qualitative recognition, this paper selects the method of LDA combined with the BP neural network after comparison. Experiments show that the qualitative recognition rate of this method in this study can reach 100%, and the amount of data after LDA dimensionality reduction is small, which speeds up the pattern speed of recognition. Regarding quantitative identification, this paper proposes a prediction experiment through Partial least squares (PLS) and BP neural networks. The experiment shows that the average relative error of the trained BP network is within 6%. Finally, the experiment of quantitative analysis of malodorous compound gas by this system shows that the maximum relative error of this method is only 4.238%. This system has higher accuracy and faster recognition speed than traditional methods.

Gesture Detection and Recognition Based on Object Detection in Complex Background

In practical human–computer interaction, a hand gesture recognition method based on improved YOLOv5 is proposed to address the problem of low recognition accuracy and slow speed with complex backgrounds. By replacing the CSP1_x module in the YOLOv5 backbone network with an efficient layer aggregation network, a richer combination of gradient paths can be obtained to improve the network’s learning and expressive capabilities and enhance recognition speed. The CBAM attention mechanism is introduced to filtering gesture features in channel and spatial dimensions, reducing various types of interference in complex background gesture images and enhancing the network’s robustness against complex backgrounds. Experimental verification was conducted on two complex background gesture datasets, EgoHands and TinyHGR, with recognition accuracies of mAP0.5:0.95 at 75.6% and 66.8%, respectively, and a recognition speed of 64 FPS for 640 × 640 input images. The results show that the proposed method can recognize gestures quickly and accurately with complex backgrounds, and has higher recognition accuracy and stronger robustness compared to YOLOv5l, YOLOv7, and other comparative algorithms.

The method of the real-time human detection and tracking

Today, data collected from video surveillance systems require processing. Video content analysis (VCA) or video analytics (VA) has found applications in security systems, retail, the automotive industry, smart home technologies, etc. The tasks of video analytics include the detection and tracking of objects. Usually, video analytics systems are specialized software for analyzing video data collected from webcams and intelligently assessing the situation. There are also separate video surveillance cameras with built-in video analytics functions. Software approaches to human detection and tracking are quite diverse, special applications and web applications or separate software modules are created. This work presents an approach to solving these problems using convolutional neural networks. The aim of the study is to increase the efficiency of human detection and tracking in video sequences. For this purpose, an overview of existing methods of detecting objects in images was conducted, in particular: the Viola-Jones algorithm, the histogram of oriented gradients. The choice of convolutional neural networks for solving the subtask of object detection is substantiated. The working principles, advantages and disadvantages of Faster R-CNN, YOLO, SSD and RetinaNet networks are considered. Their comparative analysis was carried out according to the indicators of speed and accuracy of recognition on the HABBOF dataset. A hybrid neural network for human detection and tracking has been developed: a convolutional neural network of the YOLO type has been improved. The created network achieved an accuracy of 39.2% at 43 frames per second. Experiments were carried out with the created network in order to evaluate the operation of the network in various conditions. It has been established that it works well in poor lighting conditions, but the issue of detecting small objects remains open.

Underground Mine Road Detection Using Deep Learning Technique

Semantic segmentation of underground mine roads is very important to efficiently obtain road information from images. The boundary of underground mine roads is not obvious, the environment is complex, and road identification is difficult. In order to effectively realize the accurate identification of underground mine roads, a network identification model using a deep learning technique is proposed. Choosing BiSeNet as the basic framework, adopting a unified attention fusion module, and using channel and spatial attention to enrich the fusion feature representation can effectively obtain feature information and reduce the loss of feature information. In addition, the lightweight network STDC is integrated into the backbone network to reduce computational complexity. Finally, experiments were carried out on underground mine roads. The experimental results show that the mean intersection over union and pixel accuracy of the proposed method reached 89.34% and 98.34%, respectively, and the recognition speed reached 23 f/s when identifying underground mine roads. In this study, the underground mine road recognition model trained by deep learning technology can solve the problem of underground mine road recognition with high accuracy.

SAFE: Security Door Lock System Using Haar-Cascade and LBPH Method

Door lock provides numerous benefits and has become indispensable in daily life as it acts as a security guard to prevent burglars and protect home belonging safely. The unlock methods of existing door lock system widely use keys and thumbprint involve touching the object may cause the spread of COVID-19. In this paper, a computer vision based security door lock system using Raspberry Pi (called SAFE) is proposed. Haar-Cascade classifier is employed as face detection classifier, while Local Binary Pattern Histogram (LBPH) is proposed as face recognition classifier. Recognition result is processed based on the usage of user to provide insights of SAFE. The accuracy of SAFE using pre-trained LBPH classifier achieves average of 86% based on the data obtained. The recognition speed outperforms existing work using principal component analysis and eigenfaces.

Identification of Pine Wilt Disease Infected Wood Using UAV RGB Imagery and Improved YOLOv5 Models Integrated with Attention Mechanisms

Pine wilt disease (PWD) is a great danger, due to two aspects: no effective cure and fast dissemination. One key to the prevention and treatment of pine wilt disease is the early detection of infected wood. Subsequently, appropriate treatment can be applied to limit the further spread of pine wilt disease. In this work, a UAV (Unmanned Aerial Vehicle) with a RGB (Red, Green, Blue) camera was employed as it provided high-quality images of pine trees in a timely manner. Seven flights were performed above seven sample plots in northwestern Beijing, China. Then, raw images captured by the UAV were further pre-processed, classified, annotated, and formed the research datasets. In the formal analysis, improved YOLOv5 frameworks that integrated four attention mechanism modules, i.e., SE (Squeeze-and-Excitation), CA (Coordinate Attention), ECA (Efficient Channel Attention), and CBAM (Convolutional Block Attention Module), were developed. Each of them had been shown to improve the overall identification rate of infected trees at different ranges. The CA module was found to have the best performance, with an accuracy of 92.6%, a 3.3% improvement over the original YOLOv5s model. Meanwhile, the recognition speed was improved by 20 frames/second compared to the original YOLOv5s model. The comprehensive performance could well support the need for rapid detection of pine wilt disease. The overall framework proposed by this work shows a fast response to the spread of PWD. In addition, it requires a small amount of financial resources, which determines the duplication of this method for forestry operators.

Sorting of Fresh Tea Leaf Using Deep Learning and Air Blowing

The sorting of machine-picked fresh tea leaves after mechanized harvesting remains a challenge because of the complex morphological characteristics and physicochemical properties of fresh tea leaves. First, the recognition results of four types of models, namely, YOLOv5, YOLOv3, Fast RCNN, and SSD, were compared. It was found that YOLOv5, with guaranteed recognition accuracy, had a recognition speed of 4.7 ms/frame (about four times that of the second ranked YOLOv3). Therefore, this study presents a novel fresh tea leaf sorting system that provides rapid and high-precision multi-channel sorting for four grades of tea leaves using a tea leaf recognition model based on the You Only Look Once (YOLOv5) deep learning model. Subsequently, a raw dataset, consisting of 6400 target images of different grades and different moisture contents, was used to evaluate three different optimization methods. Among these, the Stochastic Gradient Descent (SGD) optimization method was found to provide the best model training results with an average recognition accuracy of 98.2%. In addition, the recognition efficacy of the recognition model was found to be positively correlated with the gradient coverage of tea’s moisture content in the training set. Theoretical analysis was then conducted, along with the experimental investigation of the air-blowing force on the fresh tea leaves in the sorting process, with 30° determined to be the optimal air-blowing angle. Finally, the overall results showed that the construction of the full moisture content training set enabled a model recognition accuracy of up to 88.8%, a recall of 88.4%, a recognition speed of 4.7 ms/frame, and an overall sorting accuracy of 85.4%. This result is promising for multi-channel sorting of fresh tea leaf grades in complex situations, and as such provides a strong basis for the application of tea leaf sorting equipment.

Do age and language impairment affect speed of recognition for words with high and low closeness centrality within the phonological network?

Purpose Speed and accuracy of lexical access change with healthy ageing and neurodegeneration. While a word's immediate phonological neighbourhood density (i.e. words differing by a single phoneme) influences access, connectivity to all words in the phonological network (i.e. closeness centrality) may influence processing. This study aimed to investigate the effect of closeness centrality on speed and accuracy of lexical processing pre- and post- a single word-training session in healthy younger and older adults, and adults with logopenic primary progressive aphasia (lvPPA), which affects phonological processing. Method Participants included 29 young and 17 older healthy controls, and 10 adults with lvPPA. Participants received one session of word-training on words with high or low closeness centrality, using a picture-word verification task. Changes in lexical decision reaction times (RT) and accuracy were measured. Result Baseline RT was unaffected by age and accuracy was at ceiling for controls. Post-training, only young adults' RT were significantly faster. Adults with lvPPA were slower and less accurate than controls at baseline, with no training effect. Closeness centrality did not influence performance. Conclusion Absence of training effect for older adults suggests higher threshold to induce priming, possibly associated with insufficient dosage or fatigue. Implications for word-finding interventions with older adults are discussed.

Application of Optimized ORB Algorithm in Design AR Augmented Reality Technology Based on Visualization

The current media digitization and artistic strength are more powerful than the previous application. Using its advanced information display methods and technologies, this paper proposed a digital museum built by integrating digital media art with AR technology, which was helpful to analyze and solve the objective problems of current museums’ ecological imbalance and single-system function. Based on the principles and laws of augmented reality technology, the museum guide system is optimized. In the system evaluation experiment, firstly, the cultural relics of six kinds of materials are used as the target image to extract and identify the features of the image. In experiments, the recognition performance of three feature algorithms, Binary Robust Invariant Scalable Keypoints (BRISK), organizational retaliatory behavior (ORB), and Accelerated-KAZE (AKAZE), is compared. Among them, the ORB algorithm is superior to other algorithms in feature richness and recognition speed but is inferior to the other two algorithms in recognition accuracy. Therefore, this paper optimized the ORB algorithm based on the characteristics of the ORB algorithm. The ORB algorithm must calculate the orientation of the feature points before constructing the feature descriptor. After optimizing the parameters, the improved ORB algorithm not only has advantages in feature richness and recognition time but also improves the recognition accuracy up to 98.3%, which is 16% higher than the traditional ORB algorithm. Therefore, the application prospects of AR technology in digital media design are very important.

A lightweight algorithm capable of accurately identifying forest fires from UAV remote sensing imagery

Forest fires often have a devastating effect on the planet’s ecology. Accurate and rapid monitoring of forest fires has therefore become a major focus of current research. Considering that manual monitoring is often inefficient, UAV-based remote sensing fire monitoring algorithms based on deep learning are widely studied and used. In UAV monitoring, the size of the flames is very small and potentially heavily obscured by trees, so the algorithm is limited in the amount of valid information it can extract. If we were to increase the ability of the algorithm to extract valid information simply by increasing the complexity of the algorithm, then the algorithm would run much slower, ultimately reducing the value of the algorithm to the application. To achieve a breakthrough in both algorithm speed and accuracy, this manuscript proposes a two-stage recognition method that combines the novel YOLO algorithm (FireYOLO) with Real-ESRGAN. Firstly, as regards the structure of the FireYOLO algorithm, “the backbone part adopts GhostNet and introduces a dynamic convolutional structure, which im-proves the information extraction capability of the morphologically variable flame while greatly reducing the computational effort; the neck part introduces a novel cross-layer connected, two-branch Feature Pyramid Networks (FPN) structure, which greatly improves the information extraction capability of small targets and reduces the loss in the information transmission process; the head embeds the attention-guided module (ESNet) proposed in this paper, which enhances the attention capability of small targets”. Secondly, the flame region recognized by FireYOLO is input into Real-ESRGAN after a series of cropping and stitching operations to enhance the clarity, and then the enhanced image is recognized for the second time with FireYOLO, and, finally, the recognition result is overwritten back into the original image. Our experiments show that the algorithms in this paper run very well on both PC-based and embedded devices, adapting very well to situations where they are obscured by trees as well as changes in lighting. The overall recognition speed of Jeston Xavier NX is about 20.67 FPS (latency-free real-time inference), which is 21.09% higher than the AP of YOLOv5x, and are one of the best performance fire detection algorithm with excellent application prospects.

Object Detection and Recognition Techniques Based on Digital Image Processing and Traditional Machine Learning for Fruit and Vegetable Harvesting Robots: An Overview and Review

The accuracy, speed, and robustness of object detection and recognition are directly related to the harvesting efficiency, quality, and speed of fruit and vegetable harvesting robots. In order to explore the development status of object detection and recognition techniques for fruit and vegetable harvesting robots based on digital image processing and traditional machine learning, this article summarizes and analyzes some representative methods. This article also demonstrates the current challenges and future potential developments. This work aims to provide a reference for future research on object detection and recognition techniques for fruit and vegetable harvesting robots based on digital image processing and traditional machine learning.

Handwritten Digits Recognition Based on a Parallel Optoelectronic Time-Delay Reservoir Computing System

In this work, the performance of an optoelectronic time-delay reservoir computing system for performing a handwritten digit recognition task is numerically investigated, and a scheme to improve the recognition speed using multiple parallel reservoirs is proposed. By comparing four image injection methods based on a single time-delay reservoir, we find that when injecting the histograms of oriented gradient (HOG) features of the digit image, the accuracy rate (AR) is relatively high and is less affected by the offset phase. To improve the recognition speed, we construct a parallel time-delay reservoir system including multi-reservoirs, where each reservoir processes part of the HOG features of one image. Based on 6 parallel reservoirs with each reservoir possessing 100 virtual nodes, the AR can reach about 97.8%, and the reservoir processing speed can reach about 1 × 106 digits per second. Meanwhile, the parallel reservoir system shows strong robustness to the parameter mismatch between multi-reservoirs.

Convolutional Neural Network Based Defect Recognition Model for Phased Array Ultrasonic Testing Images of Electrofusion Joints

AbstractThis technical brief proposes a defect recognition model to recognize four typical defects of phased array ultrasonic testing (PA-UT) images for electrofusion (EF) joints. PA-UT has been proved to be the most feasible way to inspect defects in EF joints of polyethylene pipes. The recognition of defects in PA-UT images relies on the experience of operators, resulting in inconsistent defective detection rate and low recognition speed. The proposed recognition model was composed of an anomaly detection model and a defect detection model. The anomaly detection model recognized anomalies in PA-UT images, meeting the requirement of real-time recognition for practical inspection. The defect detection model classified and located defects in abnormal PA-UT images, achieving high accuracy of defects recognition. By comparing detection models, optimizing parameters and augmenting dataset, the anomaly detection model and defect detection model reached a good combination of accuracy and speed.

Differences in visio-spatial intelligence between non-athletes and netball players.

There is conflicting evidence regarding whether athletes have better visio-spatial skills than non-athletes. This gap may result from athletes' superiority in only some visio-spatial abilities (VSS), rather than all areas of vision. The aim of this study was to determine whether there is a significant difference in the visio-spatial intelligence between female netball players (n = 40) and non-athletes (n = 40) when comparing six visual skills (accommodation facility, saccadic eye movement, speed of recognition, peripheral awareness, hand-eye coordination, and visual memory). Following an optometric evaluation, the participants were assessed in six distinct established tests, including the hart near far rock, saccadic eye movement, evasion, accumulator, ball wall toss tests, and flash memory, to evaluate the VSS components of non-athletes and premier league netball players. For five of the six tests, there was a statistically significant (p ≤ 0.05) difference between netball players and non-athletes. Conversely, there is no concrete evidence that netball players have better visual memory than non-athletes (p = 0.277). When compared to non-athletes, netball players have significantly improved accommodation facility (p < .001), saccadic eye movements (p < .001), speed of recognition (p < .001), peripheral awareness (p < .001), and hand-eye coordination (p < .001), but not visual memory (p = 0.277). The findings that netball players perform better on a certain VSS have broad ramifications for theories of sport vision, the optimal way to choose tests, and the creation of VSS testing batteries for specific sports.

An improved lightweight network based on deep learning for grape recognition in unstructured environments

In unstructured environments, dense grape fruit growth and the presence of occlusion cause difficult recognition problems, which will seriously affect the performance of grape picking robots. To address these problems, this study improves the YOLOX-Tiny model and proposes a new grape detection model, YOLOX-RA, which can quickly and accurately identify densely growing and occluded grape bunches. The proposed YOLOX-RA model uses a 3 × 3 convolutional layer with a step size of 2 to replace the focal layer to reduce the computational burden. The CBS layer in the ResBlock_Body module of the second, third, and fourth layers of the backbone layer is removed, and the CSPLayer module is replaced by the ResBlock-M module to speed up the detection. An auxiliary network (AlNet) with the remaining network blocks was added after the ResBlock-M module to improve the detection accuracy. Two depth-separable convolutions (DSC) are used in the neck module layer to replace the normal convolution to reduce the computational cost. We evaluated the detection performance of SSD, YOLOv4 SSD, YOLOv4-Tiny, YOLO-Grape, YOLOv5-X, YOLOX-Tiny, and YOLOX-RA on a grape test set. The results show that the YOLOX-RA model has the best detection performance, achieving 88.75 % mAP, a recognition speed of 84.88 FPS, and model size of 17.53 MB. It can accurately detect densely grown and shaded grape bunches, which can effectively improve the performance of the grape picking robot.

Image segmentation for blind lanes based on improved SegNet model

The position of blind lanes must be correctly determined in order for blind people to travel safely. Aiming at the low accuracy and slow speed of traditional blind lanes image segmentation algorithms, a semantic segmentation method based on SegNet and MobileNetV3 is proposed. The main idea is to replace the coding part of the original SegNet model with the feature extraction part of MobileNetV3 and remove the pooling layer. Blind lanes images were collected through online search and self-shooting, and then the data were manually marked by LabelMe software and trained on TensorFlow deep learning framework. The experimental results show that the improved model has high segmentation accuracy and recognition speed. The pixel accuracy of blind lanes segmentation is 98.21%, the mean intersection over union is 96.29%, and the average time for processing a 416 × 416 image is 0.057 s, which meets the real-time requirements of the blind guidance system.

Optical character correction of large-curvature annular sector text in polar coordinate system

Optical character recognition (OCR) of complex morphologies represented by large-curvature annular sector text (AST) is a very challenging task. A three-segment text recognition framework consisting of detection, correction and recognition is currently an effective method for dealing with complex morphological OCR. Optical character correction (OCC) is a key component in processing largecurvature AST. This paper proposes an OCC method in the polar coordinate system, which consists of control point preprocessing, polar coordinate transformation, and image remapping. The control point preprocessing is used to normalize the control points of the large curvature AST region; the polar coordinate transformation is to convert the pixels in the rectangular coordinate system to the polar coordinate system; image remapping maps the original image in polar coordinate system to polar coordinate space for re-representation. The method proposed in this paper can be used in conjunction with most detection and recognition modules and is applicable to any language type. Furthermore, the correction process consumes very little computational resources and has little impact on the speed of text detection and recognition. Experimental results show that the proposed method outperforms state-of-the-art algorithms in large curvature AST correction and recognition experiments

A Model for Lumbar EMG Signal Recognition Based on Stacking Integration Learning

Human waist action recognition based on surface electromyography (sEMG) signals is a complex pattern recognition problem, the difficulty of which lies in the coupling of EMG signals between different actions, and a single model cannot effectively utilize the feature information to achieve high accuracy of action recognition. To this end, this article proposes a stacking integration-based EMG signal model for the waist, which introduces the stacking integration framework into EMG signal recognition for the first time, using <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${K}$ </tex-math></inline-formula> -nearest neighbor (KNN), random forest (RF)-master, XGboost, and LightGBM algorithms as base learners and decision tree (DT) model as a meta-learner to construct the integrated model. The complex feature information contained in the EMG signal is correlated with different waist movements to construct feature datasets with high correlation, and the training set is divided by using the fivefold cross-validation method to reduce the overfitting problem during the repeated training of the integrated model. Using the integrated model proposed in this article, the generalization ability of the model can be effectively improved and the recognition accuracy of EMG signals can be improved. According to experiments, the model has an average accuracy of 94.98% for the recognition of six waist movements in normal person, and the accuracy of recognition of the six movements in patients with lumbar strain was 91.5% on average, which is about 6% better than a single model, and the recognition speed can be within 150 ms, meeting the requirements of real-time recognition.

Comparison of two deep learning model YOLOF and Detectron2 for mesoscale eddies identification in the South China Sea

ABSTRACT Ocean mesoscale eddy is a special mesoscale phenomenon in the ocean, which widely exists in all oceans and marginal seas around the world. Compared with ordinary ocean circulation, mesoscale eddy has high rotating speed, strong current velocity, and average vertical depths of several kilometres, which play an important role in ocean circulation and material transport. The traditional mesoscale eddy identification and recognition have great subjectivity and usually depends on the parameters predefined or adjusted by experts, which cannot guarantee the accuracy. With the development of artificial intelligence, deep learning method has also been widely used in the detection and recognition of ocean eddies. Based on the sea-level anomaly (SLA) data provided by Copernicus Marine Environment Detection Service, this study compares the results of the ‘you only look once level feature’ (YOLOF) model and Detectron2 model in deep learning on marine mesoscale eddy detection. The mesoscale eddies in the South China Sea from 2011 to 2013 are detected and identified. The accuracy of Detectron2 model is about 57%, while the accuracy of YOLOF model can reach 90%. Comparing the performance of the two models on the same data set, it can be found that YOLOF model improves the accuracy and speed of recognition to a certain extent. The deep learning network model provides an effective technical method for the study of mesoscale eddy detection by sea surface height.

Emotion recognition in autism spectrum disorder across age groups: A cross-sectional investigation of various visual and auditory communicative domains.

Previous research on emotion processing in autism spectrum disorder (ASD) has predominantly focused on human faces and speech prosody, with little attention paid to other domains such as nonhuman faces and music. In addition, emotion processing in different domains was often examined in separate studies, making it challenging to evaluate whether emotion recognition difficulties in ASD generalize across domains and age cohorts. The present study investigated: (i) the recognition of basic emotions (angry, scared, happy, and sad) across four domains (human faces, face-like objects, speech prosody, and song) in 38 autistic and 38 neurotypical (NT) children, adolescents, and adults in a forced-choice labeling task, and (ii) the impact of pitch and visual processing profiles on this ability. Results showed similar recognition accuracy between the ASD and NT groups across age groups for all domains and emotion types, although processing speed was slower in the ASD compared to the NT group. Age-related differences were seen in both groups, which varied by emotion, domain, and performance index. Visual processing style was associated with facial emotion recognition speed and pitch perception ability with auditory emotion recognition in the NT group but not in the ASD group. These findings suggest that autistic individuals may employ different emotion processing strategies compared to NT individuals, and that emotion recognition difficulties as manifested by slower response times may result from a generalized, rather than a domain-specific underlying mechanism that governs emotion recognition processes across domains in ASD.