Fast Recognition Research Articles

Fast 3D-graph convolutional networks for skeleton-based action recognition

Research on human action recognition based on skeletons has received much attention. But most of the research focuses on improving the model’s generalization ability, while ignoring significant efficiency issues. This leads to developing heavy models with poor scalability and cost-effectiveness in practical use. This paper, we investigate the under-studied but practically critical recognition model efficiency problem. To this end, we present a new Fast Recognition Distillation (FRD) model learning strategy. Specifically, FRD trains a lightweight recognition neural network structure that can be quickly executed at a low computational cost. It can be achieved by effectively disseminating the identification probability information of the teacher network to the lightweight network. We call the probability information of the teacher network as soft-target, and FRD can learn more potential information from soft-target. In addition, we also used a particular loss function for soft-target. Through the FRD network, while basically maintaining the recognition accuracy, we minimized the network structure. Extensive experiments on the two large-scale datasets, NTU-RGBD and Kinetics-Skeleton, demonstrate that our model (FRD) is more lightweight and refined than others. Therefore, our model FRD is efficient.

A Light-Weight Artificial Neural Network for Recognition of Activities of Daily Living

Human activity recognition (HAR) is essential for the development of robots to assist humans in daily activities. HAR is required to be accurate, fast and suitable for low-cost wearable devices to ensure portable and safe assistance. Current computational methods can achieve accurate recognition results but tend to be computationally expensive, making them unsuitable for the development of wearable robots in terms of speed and processing power. This paper proposes a light-weight architecture for recognition of activities using five inertial measurement units and four goniometers attached to the lower limb. First, a systematic extraction of time-domain features from wearable sensor data is performed. Second, a small high-speed artificial neural network and line search method for cost function optimization are used for activity recognition. The proposed method is systematically validated using a large dataset composed of wearable sensor data from seven activities (sitting, standing, walking, stair ascent/descent, ramp ascent/descent) associated with eight healthy subjects. The accuracy and speed results are compared against methods commonly used for activity recognition including deep neural networks, convolutional neural networks, long short-term memory and convolutional–long short-term memory hybrid networks. The experiments demonstrate that the light-weight architecture can achieve a high recognition accuracy of 98.60%, 93.10% and 84.77% for seen data from seen subjects, unseen data from seen subjects and unseen data from unseen subjects, respectively, and an inference time of 85 μs. The results show that the proposed approach can perform accurate and fast activity recognition with a reduced computational complexity suitable for the development of portable assistive devices.

Fast Image Classification Algorithms Based on Sequential Analysis

This work is devoted to fast image recognition techniques based on statistical sequential analysis. We examine the possibility to sequentially process principal components and organize a convolutional neural network with early exits. We pay special attention to sequentially learning a multi-task lightweight neural network model in order to predict several facial attributes (age, gender, and ethnicity) based on preliminary training on the face classification task. We specifically note that the entire above-mentioned model should be fine-tuned in order to deal with the emotion recognition problem. Experimental evaluation on several datasets demonstrates that the proposed approach has high accuracy and very low running time and space complexity compared to known state-of-the-art methods.

Unit middleware for implementation of human–machine interconnection intelligent ecology construction

General speech recognition models require large capacity and strong computing power. Based on small capacity and low computing power to realize speech analysis and semantic recognition is a research area with great challenges for constructing intelligent ecology of the Internet of Things. For this purpose, we set up the unit middleware for the implementation of human–machine interconnection, namely human–machine interaction based on phonetics and semantics control for constructing intelligent ecology of the Internet of Things. First, through calculation, theoretical derivation and verification we present a kind of novel deep hybrid intelligent algorithm, which has realized speech analysis and semantic recognition. Second, it is to establish unit middleware using the embedded chip as the core on the motherboard. Third, it is to develop the important auxiliary tools writer-burner and cross-compiler. Fourth, it is to prune procedures and system, download, burn and write the algorithms and codes into the unit middleware and cross-compile. Fifth, it is to expand the functions of the motherboard, provide more components and interfaces, for example including RFID(Radio Frequency Identification, RFID), ZigBee, Wi-Fi, GPRS(General Packet Radio Services, GPRS), RS-232 serial port, USB(Universal Serial Bus, USB) interfaces and so on. Sixth, we take advantage of algorithms, software and hardware to make machines "understand" human speech and "think" and "comprehend" human intentions so as to implement human–machine interconnection, which further structure the intelligent ecology of the Internet of Things. At last, the experimental results denote that the unit middleware have very good effect, fast recognition speed, high accuracy and good stability, consequently realizing the intelligent ecology construction of the Internet of Things.

Improved EfficientDet Algorithm for Basketball Players’ Upper Limb Movement Trajectory Recognition

ABSTRACT Whether a basketball player’s movements are standard or not is a key indicator to evaluate a basketball player and a key point for basketball player training. In order to improve the training effect of basketball players, effective and fast accurate recognition of basketball players’ action trajectories is crucial. In order to solve the problems of inadequate extraction of limb action features and loss of detail information due to multiple scales of limb action features caused by traditional recognition algorithms, an action trajectory recognition algorithm based on the improved EfficientDet-D0 is proposed. The algorithm firstly adds a spatial attention mechanism in the backbone network of EfficientDet-D0, where can locate the limb action trajectory features in the image more accurately. Secondly, in the feature fusion network, in order to describe the high-frequency detail information lost by downsampling, the idea of Laplace pyramid is used to fuse the detail feature maps in the top-down fusion path, and cross-level connections are added to make full use of the feature information of different resolutions, so that the acquired high-level feature map information is richer. Finally, the whole network is trained using the migration learning technique and Adam optimizer. The experimental results show that the model achieves a PCP value (percentage correct parts) of 95.6% on the upper limb pose dataset and takes only 5.34 s for 1000 tests. Compared with the traditional algorithm, it has a higher accuracy rate and stronger robustness, with strong practical applicability, realizing the effective and accurate recognition of basketball players’ upper limb action trajectories in real time.

A novel fast pedestrian recognition algorithm based on point cloud compression and boundary extraction.

Pedestrian recognition has great practical value and is a vital step toward applying path planning and intelligent obstacle avoidance in autonomous driving. In recent years, laser radar has played an essential role in pedestrian detection and recognition in unmanned driving. More accurate high spatial dimension and high-resolution data could be obtained by building a three-dimensional point cloud. However, the point cloud data collected by laser radar is often massive and contains a lot of redundancy, which is not conducive to transmission and storage. So, the processing speed grows slow when the original point cloud data is used for recognition. On the other hand, the compression processing of many laser radar point clouds could save computing power and speed up the recognition processing. The article utilizes the fusion point cloud data from laser radar to investigate the fast pedestrian recognition algorithm. The focus is to compress the collected point cloud data based on the boundary and feature value extraction and then use the point cloud pedestrian recognition algorithm based on image mapping to detect pedestrians. This article proposes a point cloud data compression method based on feature point extraction and reduced voxel grid. The Karlsruhe Institute of Technology and Toyota Technological Institute data set is used to investigate the proposed algorithm experimentally. The outcomes indicate that the peak signal-to-noise ratio of the compression algorithm is improved by 6.02%. The recognition accuracy is improved by 16.93%, 17.2%, and 16.12%, corresponding to simple, medium, and difficult scenes, respectively, when compared with the point cloud pedestrian recognition method based on image mapping, which uses the random sampling method to compress the point cloud data. The proposed method could achieve data compression better and ensure that many feature points are retained in the compressed Point Cloud Data (PCD). Thus, the compressed PCD achieves pedestrian recognition through an image-based mapping recognition algorithm.

LCCF-Net: Lightweight contextual and channel fusion network for medical image segmentation

Accurate and fast recognition of medical images is a key step for clinical diagnosis. With the rapid development of convolutional neural networks in image processing, the convolutional neural network (CNN) approaches based on the U-shape structure have been used in various medical image recognition projects. However, the context information extraction capability of U-shape structure is insufficient and boundaries of segmentation results are blurred. In this paper, we propose a lightweight contextual and channel fusion network (LCCF-Net) for medical segmentation, which fuses multi-scale context information, preserves channel information and minimizes the amount of computation in the network. LCCF-Net is mainly composed of codec block, Shifted Windows Self-attention Connection (SWSC) block, Xception Group (XG) block and Multiple Pooling (MP) block. We design to input the features of each level of U-net encoder into their own SWSC blocks for self-attention mechanism operation, and then fuse with decoding features to participate in feature reconstruction. The proposed Multiple Pooling (MP) block combines multi-scale context information in high-level features, which can fuse rich context information. In order to reduce the number of parameters in the network, we design XG block instead of conventional convolution. Comprehensive results show that the proposed method outperforms other state-of-the-art methods for kidney tumor recognition, retinal vessel detection and COVID 19 segmentation.

YOLO-GW: Quickly and Accurately Detecting Pedestrians in a Foggy Traffic Environment.

In practice, the object detection algorithm is limited by a complex detection environment, hardware costs, computing power, and chip running memory. The performance of the detector will be greatly reduced during operation. Determining how to realize real-time, fast, and high-precision pedestrian recognition in a foggy traffic environment is a very challenging problem. To solve this problem, the dark channel de-fogging algorithm is added to the basis of the YOLOv7 algorithm, which effectively improves the de-fogging efficiency of the dark channel through the methods of down-sampling and up-sampling. In order to further improve the accuracy of the YOLOv7 object detection algorithm, the ECA module and a detection head are added to the network to improve object classification and regression. Moreover, an 864 × 864 network input size is used for model training to improve the accuracy of the object detection algorithm for pedestrian recognition. Then the combined pruning strategy was used to improve the optimized YOLOv7 detection model, and finally, the optimization algorithm YOLO-GW was obtained. Compared with YOLOv7 object detection, YOLO-GW increased Frames Per Second (FPS) by 63.08%, mean Average Precision (mAP) increased by 9.06%, parameters decreased by 97.66%, and volume decreased by 96.36%. Smaller training parameters and model space make it possible for the YOLO-GW target detection algorithm to be deployed on the chip. Through analysis and comparison of experimental data, it is concluded that YOLO-GW is more suitable for pedestrian detection in a fog environment than YOLOv7.

SheepFaceNet: A Speed-Accuracy Balanced Model for Sheep Face Recognition.

The recognition of sheep faces based on computer vision has improved the efficiency and effectiveness of individual sheep identification, providing technical support for the development of smart farming. However, current recognition models have problems such as large parameter sizes, slow recognition speed, and difficult deployment. Therefore, this paper proposes an efficient and fast basic module called Eblock and uses it to build a lightweight sheep face recognition model called SheepFaceNet, which achieves the best balance between speed and accuracy. SheepFaceNet includes two modules: SheepFaceNetDet for detection and SheepFaceNetRec for recognition. SheepFaceNetDet uses Eblock to construct the backbone network to enhance feature extraction capability and efficiency, designs a bidirectional FPN layer (BiFPN) to enhance geometric location ability, and optimizes the network structure, which affects inference speed, to achieve fast and accurate sheep face detection. SheepFaceNetRec uses Eblock to construct the feature extraction network, uses ECA channel attention to improve the effectiveness of feature extraction, and uses multi-scale feature fusion to achieve fast and accurate sheep face recognition. On our self-built sheep face dataset, SheepFaceNet recognized 387 sheep face images per second with an accuracy rate of 97.75%, achieving an advanced balance between speed and accuracy. This research is expected to further promote the application of deep-learning-based sheep face recognition methods in production.

Fast recognition model of Fusarium in agaric based on hyperspectral imaging

AbstractAgaric is a dual‐purpose fungus for medicine and food, which is considered a healthcare product. However, in cultivation and production, agaric is contaminated by a variety of molds, resulting in a decline in quality and yield. Fusarium, as one of the poisonous filamentous fungi, seriously threatens the healthy development of the agaric industry and even affects human health. Therefore, it is necessary to establish timely and accurate identification and rapid detection of pathogens and crops. In this study, the visible near infrared hyperspectral imaging was used to obtain the image and spectral information, and the best classification model was selected and established to quickly distinguish Fusarium oxysporum from Fusarium verticillioides, which was also successfully applied for the preliminary identification of agaric infection. This approach could provide a nondestructive manner for rapid identification and detection of pathogenic fungi contamination in agaric.

LSR-YOLO: A High-Precision, Lightweight Model for Sheep Face Recognition on the Mobile End.

The accurate identification of sheep is crucial for breeding, behavioral research, food quality tracking, and disease prevention on modern farms. As a result of the time-consuming, expensive, and unreliable problems of traditional sheep-identification methods, relevant studies have built sheep face recognition models to recognize sheep through facial images. However, the existing sheep face recognition models face problems such as high computational costs, large model sizes, and weak practicality. In response to the above issues, this study proposes a lightweight sheep face recognition model named LSR-YOLO. Specifically, the ShuffleNetv2 module and Ghost module were used to replace the feature extraction module in the backbone and neck of YOLOv5s to reduce floating-point operations per second (FLOPs) and parameters. In addition, the coordinated attention (CA) module was introduced into the backbone to suppress non-critical information and improve the feature extraction ability of the recognition model. We collected facial images of 63 small-tailed Han sheep to construct a sheep face dataset and further evaluate the proposed method. Compared to YOLOv5s, the FLOPs and parameters of LSR-YOLO decreased by 25.5% and 33.4%, respectively. LSR-YOLO achieved the best performance on the sheep face dataset, and the mAP@0.5 reached 97.8% when the model size was only 9.5 MB. The experimental results show that LSR-YOLO has significant advantages in recognition accuracy and model size. Finally, we integrated LSR-YOLO into mobile devices and further developed a recognition system to achieve real-time recognition. The results show that LSR-YOLO is an effective method for identifying sheep. The method has high recognition accuracy and fast recognition speed, which gives it a high application value in mobile recognition and welfare breeding.

Photo responsive dyads for fast recognition of cyanide and fluoride in aqueous media

Three new photo responsive molecular dyads have been synthesized by the reaction of 4-hydrazino-N-benzyl-1,8-naphthalimide with different diketone derivatives (benzil, acenaphthene quinone, pyrene-4,5-dione). Naph-diones have been observed to exhibit photo tautomerization plausibly due to proton transfer from the 4-amino-naphthalimide, –NH to the carbonyl, >CO moiety of the dione when photo-irradiated in a polar aprotic solvent (THF). The synthesized dyads also respond to protonation and de-protonation on change of pH of the medium and exhibited well defined multi-colour solvatochromism in different solvents. The synthesized derivatives have been determined to be highly sensitive towards critical anions (e.g., fluoride and cyanide ions). Observed photo tautomerization has been confirmed by DFT calculations and solvent induced spectral changes. The recognition of cyanide and fluoride in dilute solutions has been observed to be very fast and required less than 15 s response time. Mechanism of interaction has been confirmed through 1H NMR and UV-Vis spectra. Details of experiments conducted, energy calculations and spectral observations made for ionic recognition have been described.

Double Quantification of Template and Network for Palmprint Recognition

The outputs of deep hash network (DHN) are binary codes, so DHN has high retrieval efficiency in matching phase and can be used for high-speed palmprint recognition, which is a promising biometric modality. In this paper, the templates and network parameters are both quantized for fast and light-weight palmprint recognition. The parameters of DHN are binarized to compress the network weight and accelerate the speed. To avoid accuracy degradation caused by quantization, mutual information is leveraged to optimize the ambiguity in Hamming space to obtain a tri-valued hash code as a palmprint template. Kleene Logic’s tri-valued Hamming distance measures the dissimilarity between palmprint templates. The ablation experiments are tested on the binarization of the network parameter, and the normalization and trivialization of the deep hash output value. The sufficient experiments conducted on several contact and contactless palmprint datasets confirm the multiple advantages of our method.

Fast Localization and High Accuracy Recognition of Tire Surface Embossed Characters Based on CNN

To solve the problem of recognizing artificial tire-side pressure printing characters with low efficiency and high labor intensity, we propose a CNN-based method for tire surface character recognition. In the image pre-processing, the SSR algorithm is improved to enhance the contrast of characters, and the Normalized Cross Correlation template matching algorithm based on pyramid acceleration is proposed to quickly locate the “DOT” characters and segment them. The improved LeNet-5 network structure is used to recognize characters, and a self-built digital sample library is randomly divided according to the ratio of 8:2 to conduct digital recognition experiments. The experimental results show that the recognition accuracy of the training set can reach 95.9%, and the accuracy of the validation set is 99.5%. The accuracy of the testing set is 95.6%, which meets the practical application requirements. Moreover, the whole algorithm only needs to be implemented on a commonly configured CPU, reducing equipment costs.

The decomposition of the deep learning machine for specialized datasets for time minimizing of spatial information processing

For effective military management, the creation of complexes of automation tools, spatial information processing systems, primarily consolidated, is a priority task in the conditions of constant growth of data and requirements for their collection, transmission, storage, processing and use.
 The problem of consolidated processing of spatial information is related to the diversity of sources, the diversity of data formats, the diversity of acquisition and use, the diversity of processing etc. All that imply an extremely complex organizational and technical structure, a kind of ‘system of systems’.
 Deep Learning Machine (DLM) ensure the high accuracy of prediction. But such DLM should be matched for the military conditions of the usage where time restrictions and space lack are present.
 So, the effort to create a dataset for the maximum accuracy of the Deep Learning Machine rests on computing resources (in common research it is possible to overcome but in military sphere it isn’t). In military applied tasks, the criterion of time and overcoming uncertainties due to confrontation are critical. This allows us to put forward the hypothesis that it is impossible to achieve absolute accuracy in deep learning machine. Therefore, for variable tactical situations, it is advisable to create specialized datasets and achieve maximum efficiency from each iteration step (or their combination) using the decomposition method of the consolidated spatial information processing system. It is analysed the methods of scientific and technical solutions in the deep learning machine and the method of systematizing data types in existing tactical situations. In the end the detection and recognition system with a deep learning machine and a set of specialized datasets is proposed in the paper.
 The volume of each specialized dataset at the level of 103 enables ultra-high speed of information processing processes and allows a person to set up such a system of consolidated processing of spatial information without excessive organizational and time demands. The dataset itself is revealed.
 This principle of forming a dataset, or their sets, allows obtaining high-accurate and high -fast detection and recognition systems.

A fast recognition method for coal gangue image processing

A fast recognition method for coal gangue image processing

Manifolds-Based Low-Rank Dictionary Pair Learning for Efficient Set-Based Video Recognition

As an important research direction in image and video processing, set-based video recognition requires speed and accuracy. However, the existing static modeling methods focus on computational speed but ignore accuracy, whereas the dynamic modeling methods are higher-accuracy but ignore the computational speed. Combining these two types of methods to obtain fast and accurate recognition results remains a challenging problem. Motivated by this, in this study, a novel Manifolds-based Low-Rank Dictionary Pair Learning (MbLRDPL) method was developed for a set-based video recognition/image set classification task. Specifically, each video or image set was first modeled as a covariance matrix or linear subspace, which can be seen as a point on a Riemannian manifold. Second, the proposed MbLRDPL learned discriminative class-specific synthesis and analysis dictionaries by clearly imposing the nuclear norm on the synthesis dictionaries. The experimental results show that our method achieved the best classification accuracy (100%, 72.16%, 95%) on three datasets with the fastest computing time, reducing the errors of state-of-the-art methods (JMLC, DML, CEBSR) by 0.96–75.69%.

The bright side of CEO narcissism and its impact on accounting conservatism

PurposeThis study sheds new light on the personality trait and provides evidence regarding the relation between narcissism and desirable accounting practices, specifically the impact of CEO narcissism on accounting conservatism.Design/methodology/approachThe authors test the relation between CEO narcissism and accounting conservatism for a sample of 907 US companies and their corresponding CEOs for the period between 2010 and 2018. The authors apply three established models of accounting conservatism and measure executives' narcissism using a non-intrusive approach ubiquitous in the literature.FindingsThe authors find that CEO narcissism is associated with speculative accounting practices in the form of timely recognition of positive news and more prudent financial reporting of anticipated negative news. The authors provide the first empirical evidence that, despite its well-known negative effects on corporate financial reporting, executive narcissism can also produce positive outcomes.Originality/valueWhile managerial overconfidence has received much attention, the effects of executives' narcissism are still widely unexplored (Chatterjee and Hambrick, 2007). The authors thus contribute to the literature by investigating the relationship between CEOs' narcissism and accounting conservatism. The authors conjecture CEO narcissism should have a twofold effect on prudent financial reporting. On the one hand, CEOs' narcissism should be associated with low levels of unconditional conservatism due to excessively fast good news recognition. On the other hand, narcissistic executives should be associated with early recognition of negative news and hence with higher levels of conditional conservatism.

Application of image recognition based on neural network algorithm in multimedia dance teaching

Neural network recognition algorithm is a deep learning-based convolutional neural network, which has been applied to target detection, image recognition, sound recognition and other fields. Although the recognition technology of convolutional neural network has been gradually stabilized, it is mainly implemented in single-machine serial mode. Therefore, there will be problems such as too long training time and insufficient memory capacity. According to the previous work, this paper has carried out related research and obtained some useful results. This paper focuses on the introduction of image recognition technology based on neural network algorithm. At the same time, this paper proposes a fast real-time image recognition algorithm, which has high recognition accuracy. In this paper, the existing algorithms mentioned in the paper are analyzed, the characteristics and shortcomings of all algorithms are summarized, and the appropriate combination of algorithms is selected for specific problems. On this basis, the acceleration method of artificial neural network is used to improve the algorithm of convolutional neural network image recognition technology. With the continuous development of education and the increase of the number of students year by year, the effect of classroom teaching becomes particularly important. Therefore, modern information technology is used to transfer the classroom teaching mode to the network and realize a new method of online remote online dance teaching. According to the actual situation of dance teaching, the advantages of multimedia dance teaching are analyzed. The software of multimedia dance teaching system is tested comprehensively, from the aspects of computer system memory occupancy rate and temperature, to test whether the activities of B/S mode multimedia dance teaching system meet the expected goals and requirements of each core module of multimedia dance teaching system.

A Machine Learning Methodology for Diagnosing Chronic Kidney Disease

Chronic kidney disease (CKD) is a global health problem with high morbidity and mortality rate, and it induces other diseases. Since there are no obvious symptoms during the early stages of CKD, patients often fail to notice the disease. Early detection of CKD enables patients to receive timely treatment to ameliorate the progression of this disease. Machine learning models can effectively aid clinicians achieve this goal due to their fast and accurate recognition performance. In this study, we propose a machine learning methodology for diagnosing CKD. The CKD data set was obtained from the University of California Irvine (UCI) machine learning repository, which has a large number of missing values. KNN imputation was used to fill in the missing values, which selects several complete samples with the most similar measurements to process the missing data for each incomplete sample. Missing values are usually seen in real-life medical situations because patients may miss some measurements for various reasons. After effectively filling out the incomplete data set, six machine learning algorithms (logistic regression, random forest, support vector machine, k-nearest neighbor, naive Bayes classifier and feed forward neural network) were used to establish models. Among these machine learning models, random forest achieved the best performance with 99.75% diagnosis accuracy. By analyzing the misjudgments generated by the established models, we proposed an integrated model that combines logistic regression and random forest by using perceptron, which could achieve an average accuracy of 99.83% after ten times of simulation. Hence, we speculated that this methodology could be applicable to more complicated clinical data for disease diagnosis.