In order to improve the recognition performance and accuracy of traditional object detection and to accelerate the computation speed, a CNN( Convolutional Neural Network) model with more powerful feature learning and representation capabilities and with related deep learning training algorithms is adopted and applied to large-scale recognition tasks in the field of computer vision. The characteristics of traditional object detection algorithms, such as the V-J(Viola-Jones) detector, HOG(Histogram of Oriented Gradients) features combined with SVM( Support Vector Machine) classifier, and DPM ( Deformable Parts Model) detector are analyzed. Subsequently, the deep learning algorithms that emerged after 2013, such as the RCNN ( Region-based Convolutional Neural Networks) algorithm and YOLO(You Only Look Once) algorithm are introduced, and their application status in object detection tasks is analyzed. To detect fallen individuals, the YOLOv5(You Only Look Once version 5) model is used to train the behavior of individuals with different heights and body types. By using evaluation metrics such as IoU(Intersection over Union), Precision, Recall, and PR curves, the YOLOv5 model is analyzed and evaluated for its performance in detecting both standing and fallen activities. In addition, by pre- training and data augmentation, the number of training samples is increased, and the recognition accuracy of the network is improved. The experimental results show that the recognition rate of fallen individuals reaches 86% . The achievements of this study will be applied to the design of disaster detection and rescue robots, assisting in the identification and classification of injured individuals who have fallen, and improving the efficiency of disaster area rescue.

In order to solve the program code similarity detection measurement which ignores the program semantics and the invalid measurement, we present
 an AST(Abstract Syntax Tree) based on the program code similarity measure method. Through the pretreatment redundancy removal in AST generation and the lexical grammar analysis, get the corresponding AST; and then according to the adaptive threshold method,using the AST traversal which include the sequence and process attributes to take the similarity calculation;finally,determine whether plagiarism and generate the test report.The experimental results show that this method can effectively detect a variety of plagiarism java code.

In order to improve the search speed and shorten the search time of robot path planning，the characteristics of various algorithms is summarized． First，the history of mobile robot development and outline the key technologies of path planning are reviewed． Secondly，the mobile robot path planning is classified and summarized． From the perspective of the mobile robot's grasp of the environment，the mobile robot path planning is divided into two categories: global planning and local planning． Then the related algorithms of global planning and local planning are reviewed，and the development status，advantages and disadvantages of related algorithms are summarized． Finally，the future development trend of robot path planning technology in further research，hybrid algorithm， multi-robot collaboration， complex environment and multi-dimensional environment is pointed out．

To meet the need of integrating scientific research into teaching of Emerging Engineering Education, a multi-dimensional and hierarchical integrated experimental platform based on Python is designed. Guided by the talent-training plan, hierarchical modules involving image recognition, machine learning and data analysis is designed from scientific research hotspots. Image recognition module starts from character recognition, then face and license plate recognition are realized by several algorithms. In the machine learning module, commonly used machine learning algorithms are studied and corn disease is identified by various methods based on Python. In the data processing and analysis module, Excel data processing experiment based on Python is designed to analyze the data of workload and bioinformatics data. The platform enables students to learn the application of Python in the experiments, and choose different experimental projects according to professional needs and research directions to realize the goal of teaching students in accordance with their aptitude. By applying the experimental platform to teaching practice, it is demonstrated that students have a deeper understanding of Python’s programming implementation in image recognition, machine learning, and data analysis and enhanced research interest. And the goal of integrating scientific research into teaching and improving the quality of undergraduate teaching could be achieved.

In order to promote the development of path planning technology，the planning ideas and forms of path planning are analyzed． The path planning algorithms are divided into the traditional classical algorithms and modern intelligent algorithms in two categories，and some commonly used algorithms are analyzed and summarized．And the current research hotspots and future development trends are pointed out from the three aspects of improving the application of modern intelligent algorithms in path planning，amalgamation of multiple algorithms and the research of four-dimensional path planning algorithms for multiple UAVs．

The process of behavior recognition can be regarded as the combination of feature extraction and
classifier to a large extent． Compared to static image object recognition，video feature extraction of human
behavior recognition is more susceptible to such factors as dynamic background，acquisition device motion，
perspective and illumination，so it poses great challenges to researchers． Based on the systematic classification of
behavior recognition feature extraction，according to the different types of behavior recognition feature extraction
methods and common behavior recognition database，the behavior recognition feature extraction is systematically
classified to expound the latest research progress． And the current research difficulties and possible future
research directions are discussed．

In order to solve the human intervention problem of existing human target recognition and tracking
system，an automatic human detection and tracking system is designed，which is composed of embedded system，
wireless communication technology and upper computer． The automatic detection and tracking system is divided
into two parts: the upper computer and the lower computer． Using STM32F103ＲCT6 as the control unit，the
lower computer detects the position of the human body through the SHＲAP-GP2Y0A21YK0F infrared ranging
sensor，and then controls the steering gear． The steering gear is equipped with a camera to collect the video
signal which is transmitted to the upper computer using WIFI( Wireless Fidelity) wireless technology． The upper
computer is developed by Eclipse-Android system development platform，which can display the video signal in
real time in the monitoring center． It has been proved that the system is easy to install and operate，can
accurately realize the automatic recognition and tracking process of human body，and can be widely used in the
industry of indoor non-interference infrared work．

 To enharce the service quality of the internet catering industy, to better meet the needs of customers at ang time pointmed, the research is an intelligent ordering system based on Android Studio development environment, C/ S structure, and Gradle and Genymotion development tools. This paper introduces the principle of image loading and the interaction process between activities and the background. The system has realized its functions such as menu browsing, commodity collecting, food ordering, user center and food choosing by shaking. The result of tests show that the system has a good user interaction experience and can be operated easily. And it will have an extensive application prospect.

In recent years, benefiting from the mature application of communication, big data, cloud computing and other technologies, “ Internet + ” has been widely popularized in people’s livelihood, economy and government affairs. With the rapid increase of equipment and explosive growth of data, the network environment becomes increasingly complex and brings huge hidden dangers to network security. The need for the development of network security technology and how to apply artificial intelligence AI(Artificial Intelligence) to help solve some problems is introduced. And how to apply machine learning ML(Machine Learning) to improve network security performance is analyzed in detail for specific domain or specific network technology. First, we summarize existing research work on using AI to combat cyber attacks, including using traditional machine learning approaches and existing deep learning solutions. We then analyzed the counterattacks that the AI itself might be subjected to, dissected their characteristics, and classified the appropriate defenses. We also provide some advanced concepts of artificial intelligence network security technology i. e. how to better apply artificial intelligence to the field of network security in the future.

 Because the request of automobile safety is higher and higher, drivers should clearly ensure the surrounding environment of the car, a kind of car surveying technology is proposed. Video mosaicing technology is the basis of surveying technology. In order to meet the requirement of vehicle instantaneity, this paper uses FPGA to achieve the video mosaicing. The system includes video capture module, video pre-processing module, video mosaicing module and video display module. The paper uses Cyclone IV development board of Altera company to complete the system and uses double DDR2 to complete Ping-Pong buffer. We use sift algorithm to match videos and linear fusion algorithm to mosaic videos. The system is verified by experiments and its feasibility is proved.

As crowd-sensing networks can efficiently address the key problem of large-scale sensor networks,i. e. large quantities of deployment and maintenance cost, it becomes a research hotspot for the Internet of Things. This paper presents the state-of-the-art of crowd-sensing networks. We first introduce the definition, the origin and the framework of crowd-sensing networks, followed by the analysis of its characteristics and the comparison with traditional sensing networks. And then, recent researches of crowd-sensing networks are surveyed in terms of environment sensing, public infrastructure sensing and social sensing. Finally, we point out three important directions for the future research of Crowd-Sensing Networks and analyze their challenges, to give a reference and help for the future study.

In order to solve the problems of efficient storage, processing and analysis of medical data, a medical big data platform is designed and developed. HDFS (Hadoop distributed file system) is built and deployed, and a website platform based on Tomcat server is designed. The web server is combined with distributed file system by writing Hadoop web API, and a Python script program with high data processing efficiency is designed to read and analyze medical data. The test results show that medical big data platform achieves the expected functions of data storage, sharing and visualization.

For the needs of three-dimensional geological project model display, a three-dimensional model network sharing methods based on Unity3D is presented. We have designed and implemented major functions, such as picking up objects and geological information queries, hiding and showing geological objects, real time cross sectioncutting and model blasting. The system is focused on the three-dimensional scale display. It is convenient for researchers to obtain information in the deep geological and achieved good results.

To ensure the authenticity and information safety of the image content, application areas of passive forensics techniques are stated. The ideas and main steps of some typical algorithms for digital image source device detection, splicing detection, copypaste detection, blur detection and JPEG （Joint
Photographic Experts Group）compression detection have been analyzed. The algorithm performance and feasibility have been briefly evaluated, and the future technical route and development direction of passive forensics techniques are prospected.

In order to realize the intelligent perception oriented CNC (Computer Numerical Control) system，
based on the characteristics of the network communication and open architecture of the intelligent CNC system，
we design a secondary development platform of the intelligent CNC system and the reconfigurable hardware
structure of the system，and put forward the multi-channel signal processing technology，which ensures the real-
time processing，reduces the interference between the signals and improves the processing efficiency． In order to
satisfy the error caused by nonlinearity and time delay in CNC system，the adaptive control of servo motor of CNC
machine tool is realized based on the improved MFAC(Model-Free Adaptive Control) data driving algorithm．
With the SSB field bus developed by ourselves to collect sensor signals，the design of data frame format and
communication protocol is completed to ensure the low power consumption and high reliability of the information
acquisition system． The improved MFAC data-driven algorithm is tested by the test and verification platform
based on blue sky CNC，and good test results are achieved．

Under the current situation of normalized management of epidemic situation, in order to ensure the normal development of computer experiment courses in colleges and universities, a virtual laboratory for computer basic experiments is established based on Docker technology. Students can access the server through a browser to obtain an independent experimental environment. The Docker-Compose tool is used to create, open, stop, delete and other multi-dimensional management of students' experimental environments, and to ensure their performance, which is equivalent to moving the offline laboratories online. This scheme can meet the needs of online computer basic experiments and provide high-quality experimental services for corresponding theoretical teaching.

 In response to the problem of many class hours and many students attending classes in public computer laboratories each semester, a set of public computer laboratory information system has been developed independently. The system consists of three parts: student side, teacher side and server. The student side is a desktop program based on Python. The teacher side and the server are implemented in a web project written by JSP(Java Server Pages). In terms of function, the platform can be divided into three basic modules: student sign in, lost and found, feedback. It integrates other common functions. The application results indicate that this system can utilize information technology to provide convenience for students to learn in the laboratory. It truly implements the teaching concept of “student-oriented”

The clinical method of blood glucose measurement at present is extracting blood from the patients, which is based on the principle of biochemical reaction. This invasive methord is painful, susceptible and discontinuous. To deal with these problems, we focus on a non-invasive way on the strength of the photoplethysmography detection theory by irradiating the finger with two beams of near infrared lights of 805 nm and 940 nm. The wave of 805 nm is sensitive to glucose, while the 940 nm one acts as the reference. By processing the information of the blood glucose extracted from the transmitted lights, the blood glucose value can be calculated.Experiments have convinced that the continuous non-invasive blood glucose measurement can be achieved.

According to the image fusion of optical and SAR(Synthetic Aperture Radar), the multimodal and multiscale features including pixel features, texture features and edge features were analyzed in order to improve the traditional homologous image registration and fusion algorithm. Then the improved SURF(Speeded Up Robust Features) operator, texture analysis and contour extraction algorithm were adopted to obtain the multimodal and multiscale features of the heterologous images. By standardization algorithm of the fuzzy scale and dimension, the differences between the feature pairs of the heterologous images were overcome, which made the matching of the feature pairs available. The accuracy of registration and fusion were ensured through the method of fuzzy
correlation coefficient, and the registration and fusion of optical-SAR images were completed. Finally, the modified algorithm was verified and compared with the traditional fusion methods. Experimental results show that the multimodal registration and fusion algorithm can improve the precision and adaptability of optical-SAR registration. The average accuracy rate of registration and fusion can reach to 87. 7%, which can satisfy the requirement of high precision registration and fusion.

Taget Detection of Photovoltatic Remote Sensing Based on Improved Yolov5 Model TONG Xifeng, DU Xin, WANG Zhibao (School of Computer and Information Technology, Northeast Petroleum University, Daqing 163318, China) Abstract: Aiming at high-sensing photovoltaic image resolution, high environmental noise, and complex background, an improved Yolov5 model is proposed to achieve positioning of photovoltaic power plants. First of all, the CA(Coordinate Attention) mechanism is added to the compassionate layer of the main feature extraction network to improve the learning ability of the network characteristics; second, the Ghostconv network structure is added to Backbone, useing the Ghostconv network module to replace the Conv network module, designing a new GhostC3 network network instead of the original C3 network module to improve the learning efficiency of the model; finally, the GIoU_Loss function is changed to the SIoU_Loss function. Compared with the original Yolov5 method, the average accuracy of the improved algorithm mAP, accuracy, and recall rate reached 97. 5% , 98. 9% , and 94. 9% , respectively, which have increased by 1. 8% , 1. 7% , and 5. 8% , respectively. The algorithm has a good effect on photovoltaic detection.

In order to minimize the huge economic losses to people’s living and the country caused by major
natural disasters and to establish an effective disaster response transportation system，we proposed modern
intelligent algorithms such as clustering model based on K-means and multi-person short-circuit model based on
genetic algorithm，and carried out disaster rescue simulation experiment combining with Puerto Ｒico’s urban
data，which improved the traditional low traffic disaster response system limitation，pertinence and low efficiency
of faults． And for hospitals，road network density，population density and plains the model will provide the first
aid． We use drones to quickly inspect the main traffic lines and quickly restore the traffic． Taking Puerto Ｒico as
an example，the experiment results show that the model quickly realizes the traffic recovery，provides great
convenience for the transportation of ground materials，and improves the rescue speed．

The grid connection of electric vehicles and distributed power supplies brings significant uncertainty to the grid． In order to make grid analysis closer to the actual grid． It simulates output through the probability density function of load and distributed power output． It proposes to use Latin Hypercube Monte Carlo simulation and radial basis neural network to deal with distributed power supply and electric vehicle probability model． This method takes full account of the randomness，intermittency and correlation of EVs and distributed power supplies． Using Latin hypercube MCS ( Monte Carlo Simulations) to compare with traditional Monte Carlo simulation methods，the sampling scale is reduced and sampling coverage is improved． The radial basis neural network is used to solve the power flow calculation equation，which avoids the calculation of Jacobian matrix and partial guide in the traditional method． Through simulation，the calculation results of the proposed algorithm in the improved IEEE14 and IEEE118 node systems show that while ensuring accuracy，the algorithm speeds up the algorithm greatly，and is suitable for the solution of probabilistic power flow in large-scale power systems． The improved IEEE118 node in the system，the running time is reduced by 99． 9% compared to the traditional MCS．

In order to improve the precision of remote sensing image for crop prediction and the efficiency of agricultural planting, combined with the innovation and entrepreneurship training program of Jilin University, an experimental project of crop classification algorithm based on satellite remote sensing image is designed. Taking the high-resolution satellite image of Harbin agricultural demonstration base captured by sentinel-2 on July 30, 2018 as the experimental data, the characteristics of rice, soybean, corn and sorghum in the image are extracted and classified by using the maximum likelihood method, support vector machine method and neural network method in different spectral bands ( including red band) , and then the crop classification map is obtained, the statistical results are compared with the real parameters, and then the classification accuracy and reliability of different algorithms are compared. The experimental results show that the neural network method has the highest accuracy and the strongest reliability, and is suitable for nationwide promotion.

There are many defects in the mobility and convenience of the handwritten digit recognition system running on the computer. In order to make improvement for these defects, a handwritten digit recognition system based on the visual library OpenCV is designed, which transplants the digit recognition algorithm into the flexible and small high-performance embedded equipment. By adjusting the shooting Angle of the steering gear and using the technology of picture splicing and digital segmentation, the handwritten digit recognition of short distance and large area is realized. The recognition speed, recognition accuracy and model volume of the models trained by KNN(K-Nearest Neighbor), support vector machine and artificial neural network are compared. After testing, the identification time of Raspberry Pi by using the artificial neural network algorithm can be as low as 0. 115 s, and the recognition accuracy can reach 72% , which has a certain application value.

Abstract: To complete the ECG signal feature points extraction and the classification of ST segment, we put forward an algorithm based on the discrete wavelet transform, combined with the f derivative and the SVM(Support Vector Machine). The algorithm can accomplish the signal preprocessing, noise elimination, QRS complex detection and extraction of characteristic value, calculating the average ST segment, curve area and the standard deviation, and the simple classification of ST segment by using the SVM combined with the three sets of data. The matlab simulation results show that the wavelet denoising is effective and has no distortion, and completely extract ST segment feature points. The data are downloaded from the MIT-BIT database, the classification results show that cross-validation average accuracy is 80. 70%,the average accuracy of training is 91. 83%, the average testing accuracy was 74. 28%.

In order to reduce the inspection burden of electric power workers and realize intelligent inspection in substation, the algorithm of substation equipment defect detection is studied. Firstly, the data augmentation method is used to expand the initial dataset and various image processing method is used to generate the dataset with complex illumination environment. Then, the adaptive spatial feature fusion method is used to mitigate the inconsistency of different scale features in the feature pyramid, and the loss function of confidence is changed to Focal loss function to mitigate the imbalance between positive and negative samples. Based on the improved YOLOX-s(You Only Look Once X-s) network model, the algorithm of substation defect detection is designed. Finally, the detection effect of the improved YOLOX-s model is compared with that of other deep learning algorithms. Under the designed data set, the experiment shows that the comprehensive detection effect of the improved YOLOX-s network model is good, and the accuracy and real-time performance is satisfied. 

A football competition risk intelligent warning method based on an improved Copula model is proposed to address the issues of large errors between warning values and actual values, and multiple false alarms in football matches. Based on the fuzzy comprehensive evaluation matrix, the evaluation system for football competition risk indicators is determined. The indicator level status is classified, the Copula function is selected, and an improved Copula football competition risk intelligent warning method is constructed to accurately judge football competition risks and reduce risk losses. The experimental results show that the interference suppression of this method is high, maintained above 20 dB, and have high anti-interference ability. It can effectively suppress interference. This method also reduces the error between the warning value and the actual value, reduces the number of false alarms in the warning, and verifies the practicality and feasibility of this method.

To avoid the gradient vanishing or exploding issue in the RNN(Recurrent Neural Network)-based sequential recommenders, a gated recurrent unit based sequential recommender DeepGRU is proposed which introduces the residual connection, layer normalization and feed forward neural network. The proposed algorithm is verified on three public datasets, and the experimental results show that DeepGRU has superior recommendation performance over several state-of-the-art sequential recommenders ( averagely improved by 8. 68% ) over all compared metrics. The ablation study verifies the effectiveness of the introduced residual connection, layer normalization and feedforward layer. It is empirically demonstrated that DeepGRU effectively alleviates the unstable training issue when dealing with long sequences. 

In order to solve the problem that using traditional technologies such as Java and .NET to develop and deploy data services are complex and difficult to integrate with advanced cloud and container technologies, a lightweight 3D visualization data service solution for drilling based on Web is proposed, providing data interface support for front-end visualization applications. Based on NodeJS、 Angular、 TypeScript and other open source lightweight technologies, a lightweight drilling database system is designed, which can be used as an auxiliary tool for front-line technical managers and providing the most concerned data items in the fastest way with high efficiency and practicability. With the data loading tool, drilling technicians can easily load data into the database, including surface and seismic slices, measurements, events and well logs of blocks. And the system provides a comprehensive data security mechanism, including JWT ( JSON Web Token ) based identity authentication and JWE ( JSON Web Encripytion ) based data encryption, to ensure data security. The application results show that this solution can provide efficient data transmission services for drilling 3D visualization systems. 

In order to achieve to detect a variety of physiological signals form ECG ( electrocarcliogram)monitoring signal, reducing the complexity of the monitoring equipment, according to the impact of breathing motion on the ECG, it presents a respiratory information extracted from the ECG (ECG-derived respiratory signal, EDR) algorithm. First, according to Pan & Tompkins algorithm extracted from the ECG R-wave, and then, using a natural cubic spline interpolation algorithm to estimate the baseline ECG and remove baseline wander, getting clean ECG, finally, get clean ECG R-wave, the use of natural spline interpolation algorithm is applied to obtain R wave amplitude modulation signal which contains respiration information. In paper, it selected Fantasia database, which in addition to providing the ECG signal, it also gave a breathing signal synchronized recording. It used Matlab software to verify the proposed algorithm, associated analyzed and compared. The simulation results showed that the algorithm mentioned in this paper can extract the respiratory
signal from ECG. By comparing with respiratory data in the database simultaneously recorded, it confirmed the validity of the algorithm.

To understand and dig clinical information of CT images, and to study the method of 3D reconstructionand visualization of CT image, we present an algorithm of 3D reconstruction. A group of 16 row spiral CT images of the abdomen of the human body was imported into Mimics 10.01 using the tools thresholding, mask editing, region growing and reconstruction. Spleen was separated from the whole abdomen CT images and reconstructed. It can be o
bserved from any perspective, zoomed in and out. The experiment results show that by using proposed 3D reconstruction algorithm, the spleen reconstructed is closer to its real 3D structure. It can provide further data in detail for medical research.

To solve the problem of semiconductor doping concentrations，it is necessary to test the semiconductor in a simple and easy way． Four-probe method is a common method of measuring resistivity in the field of microelectronics technology． Combining the construction of the semiconductor device physics and experiment course，the experiment of testing the doping concentration of semiconductor by four-probe method was developed． Through the establishment of two theoretical models of semi-infinite sample model and infinite thin-layer sample model，the resistivity test methods for semiconductor materials with different thicknesses are studied，and the principle is discussed． In order to solve the problem that the commercial equipment is expensive and can not meet the needs of experimental，the author proposed to build a test system，using a simple manual probe station of tungsten alloy，and designed a four-probe test device according to the experimental needs． The practical application shows that the establishment of the four-probe test system meets the teaching requirements of semiconductor physics experiments．

Traditional cross-language information retrieval has problems such as low translation mapping
accuracy and semantic deviation after query expansion． To deal with this problem，a method of integrating
statistics and ontology is proposed to construct a multi-language information retrieval model． Using statistical
translation to solve the problem of translation mapping ambiguity，the multi-ontology BabelNet is used to
reduce the loss of semantic relevance． Because the ontology contains a large number of conceptual
connections，the ontology is used as the semantic layer representation to design the semantic weighting
algorithm． And it is built on the BM25F statistical information retrieval model as the user feedback sorting
algorithm． Finally，the multi-language information retrieval prototype system is designed according to the
established model，and the model is tested with the data set obtained based on the crawler technology． The
experimental results show that the average precision of the model is higher than the traditional machine
translation-based information retrieval model．