126results about How to "Address subjectivity" patented technology

An embodiment of the invention discloses car using type prediction model establishment and information providing methods and devices. The car using type prediction model establishment method includes the steps of obtaining history car using order information of at least two users; extracting the car using scene information associated with the history car using order information; and generating a car using type prediction model by training a set prediction model based on the car using type and the car using scene information corresponding to the history car using order information. The technical scheme in the embodiment of the invention can solve the technical problems of high cost, poor timeliness and strong subjectivity in the prediction of the car using type in the prior art, and makes full use of the history car using order information that can reflect the actual needs of using cars of the users to optimize the car using type prediction technology and improve the accuracy of pushing the car using service information of the users.

The invention discloses a self-coding network and a training method thereof, and a method and a system for detecting abnormal power consumption, wherein the training method comprises the following steps: splicing the sample power data by a sliding window to obtain a training sample set, and marking the training sample set containing the surveyed user to obtain a labeled sample, wherein the unlabeled training sample is a non-labeled sample; carrying out unsupervised training on the self-coding network by unlabeled samples, and obtaining the initialization parameters of the self-coding network,then taking the discrete class tags obtained from the coding layer of the coding network as classifiers, carrying out supervised training on the classifiers by the labeled samples, and updating the parameters of the coding layer to obtain the trained self-coding network; then, utilizing the trained self-coding network to detect the power data of the user to be measured, and judging whether the user to be measured abnormally uses power. The invention can mine abnormal information in low-density electric power data, avoid noise data interference, and improve abnormal detection accuracy.

The invention relates to a method for detecting and classifying pixel defects in inkjet printing of a printed display panel and a device thereof. The method comprises the following steps: obtaining the image of a panel to be tested in blocks; when an original image is received, preprocessing the original image; when the pre-processed original image is received, stitching all the partitioned imagesof the same panel ; when the original image is received, segmenting the image is segmented; when the segmented original image is received, using a classifier in machine learning to classify the image, if there is no defect, the image entering the subsequent process, and if there is any, giving the prompt; when the original image of the defective pixel is prompted after the classification is received, positioning and subclassifying the defective pixel of the image, and carrying out statistical analysis of the category and the number of the defective pixels of the image to indicate corresponding defects. The invention aims at solving the problems that the manual detection and classification exist for a long time, the efficiency is low, the detection result is easy to be influenced by humansubjective factors, and the consistency is poor.

The invention discloses a sunflower disease recognition method based on a random forest method. The sunflower disease recognition method can recognize four common diseases of sunflower leaves including powdery mildew, bacterial leaf spot, black spot, and downy mildew and comprises steps of: A: disease image acquisition in which an acquired leaf image color is required to be as close as possible to the color of a leaf itself; B: disease image processing in which a processing method suitable for sunflower disease image recognition is used; C: disease image segmentation in which an optimal color image segmentation method is selected by analyzing and comparing various image segmentation methods; D: disease image feature extraction in which parameters such as the color feature and the texture feature of the disease image are extracted for research; and E: disease recognition and diagnosis in which the sunflower disease is diagnosed and recognized by using the random forest method. The sunflower disease recognition method mainly solves the subjectivity, the limitation, and the fuzziness of eye determination and difficulty in new disease determination in a process of disease recognition, improves the accuracy of diseases recognition, and provides agricultural workers with good help for recognizing and preventing sunflower diseases.

The invention provides a land surface temperature-vegetation index feature space dry and wet limit selecting method and device. The method includes the steps of firstly, determining the size of a research area, and performing pretreatments such as projection conversion, resampling and cutting on satellite remote sensing image data; secondly, removing image metadata influenced by cloud and terrain; thirdly, using the rest image metadata to build a land surface temperature-vegetation index two-dimensional scatter diagram feature space; fourthly, sampling the highest land surface temperature corresponding to different vegetation index conditions obtained in the third step; fifthly, linear fitting the highest land surface temperature and vegetation indexes to obtain the dry limit of the feature space through threshold setting and an iteration algorithm; presuming that the wet limit temperature is a constant and equals to the dry limit temperature at the total vegetation cover to obtain the wet limit temperature. By the arrangement, the problems that the current land surface temperature-vegetation index feature space dry and wet limit selecting is high in subjective randomness and uncertainty, and the like are solved.

The invention relates to a cigarette smoke acidity index evaluation model establishment method based on gustation vitality values, and belongs to the technical field of cigarette evaluation. The method comprises the steps as follows: a gas chromatograph-mass spectrometer is utilized to detect the content of organic acids in mainstream cigarette smoke with an N,O- bis(trimethylsilyl)trifluoroacetamide derivatization method; according to a public reported gustation threshold, the gustation vitality values of the organic acids are calculated so as to recognize key acidity substances in the mainstream cigarette smoke; a cigarette smoke acidity index evaluation model is established on the basis of the key acidity substances in the mainstream cigarette smoke; and human sense evaluation of the cigarette smoke acidity intensity is performed by a cigarette sense technician, and statistical software is adopted to perform fitting effect analysis on the cigarette smoke acidity index and a human sense evaluation value. The cigarette smoke acidity index evaluation model and a human sense evaluation method have an ideal fitting effect, the cigarette smoke acidity intensity is objectively evaluated, and the method can be used for development of the cigarette product flavor technology.

The invention discloses a human motion posture recognition and evaluation method, and the method comprises the following steps: S01, collecting a video image test data set, and performing data processing on data in the video image test data set; S02, inputting the test data after data processing into the trained LSTM neural network model for human motion posture recognition, and outputting a recognition result; S03, comparing the output identification result with standard motion data, and evaluating the standard degree of the identified human motion posture according to the comparison result.The invention further discloses a human motion posture recognition and evaluation system, and the system comprises an image acquisition module, a data processing module, an LSTM neural network model,a model training module, a data center and a posture evaluation module. According to the method, the long-term time information can be captured, complex space-time clues of human body actions can be captured through a time domain attention mechanism and a space domain attention mechanism, and the accuracy of human body motion posture recognition is greatly improved.

The invention discloses a blast furnace smelting expert system built based on a pattern recognition technology, which comprises a server, an operator workstation, six sets of PLCs (programmable logic controllers) (respectively used for material preparation below grooves, material distribution at the top of a blast furnace, water granulated slag control, dust removal control, coal powder injection control and water system control), a distributed control system, an IPC (industrial personal computer) for communication, a state classification module, a state diagnosis module, a state evaluation module, a state adjusting module and a state display module, wherein the server is provided with a database, the state classification module, the state diagnosis module, the state evaluation module and the state adjusting module are arranged in the server with the database, and the state display module is arranged in the operator workstation. Meanwhile, the invention also discloses a method for building a blast furnace smelting expert system based on a pattern recognition technology, which comprises the following steps of data access, state classification, state diagnosis, state evaluation, state adjustment, state display and execution. The system and method disclosed by the invention have the characteristics of standard and objective evaluation, good integrated performance, rich knowledge base and real-time updating, and can be widely applied to the technical field of blast furnace process control.

The present invention discloses a device and method of on-line automatic count of fries on the basis of machine vision. The device provided by the invention comprises a red LED backlight, an industrial camera, pipelines, a filter screen, a round plane, an uncovered cylindrical sink, electromagnetic valves, a liquid level meter, a PC machine containing an image acquisition card, a single-chip microcomputer controller and a support. The acquired images are subjected to arithmetic processing such as grey level transformation, Laplace filtering, adaptive median filtering, OTSU threshold segmentation, morphological processing, endpoint refinement and the like and then fry counting through necessary software. The device provided by the invention is able to be continuously and effectively used for a long time under the control of the PC machine, so that deficiencies of traditional manual counting, such as limited working time, low precision, obvious subjective factors and the like, are solved; and the water level may be detected and controlled through the liquid level meter to adapt counting of different fries with different sizes so that different users' counting requirement may be satisfied.

The invention relates to a component distribution method for an automobile seat framework. The component distribution method for the automobile seat framework comprises the steps that a design model is established according to the design request of the seat framework, and a design area is determined; a finite element model of the automobile seat framework is established according to the design area and loading conditions; topological optimization design is conducted on the finite element model with a variable density method, and a material density distribution diagram in the design area is obtained; pipe structural parts of the automobile seat framework are arranged based on the material density distribution diagram. The reasonable material density distribution diagram is obtained by establishing the finite element model for the design area, the pipe structural parts are distributed and arranged based on a material density method, and the purposes that under the premise that the safety of the automobile seat framework is met, the weight of the pipe structural parts is reduced, and the rigidity and the strength of the pipe structural parts are improved are achieved. The lightweight design of the automobile seat framework is achieved, and good economical efficiency can be obtained.

The embodiment of the invention discloses an urban group boundary identification method and device, equipment and a storage medium. The method includes the steps of acquiring interesting points of a target area for nuclear density analysis; generating at least two contour lines according to results of the nuclear density analysis, and determining a density value corresponding to each contour line;calculating the sum of areas corresponding to the same density value, and constructing a curve function of the sum of the density values and the areas; determining a critical value range of boundaries of an urban group according to the curve function. According to the embodiment, the practical problems that the boundaries of national and regional identification urban groups are high in subjectivity and methods are complicated or lack theoretical bases are solved, and the boundary range of the urban group can be flexibly and accurately identified in real time according to distribution conditions of the spatial interesting points of actual human activities.

The invention discloses a parameter for reflecting biological tissue abnormity and a measuring method thereof, and belongs to the technical field of hypothesis testing in B-mode ultrasound diagnosis and statistics. The method comprises the steps of firstly setting a B-mode ultrasound diagnosis apparatus into a second harmonic scanning mode to scan and store original RF data of tissue; secondly conducting band-pass filtering on one frame of the RF data to select interested and reference areas in the frame of the RF data; then calculating the root-mean-square value of second harmonics in each cycle or several cycles on each scanning line in the interested area and reference area; then conducting Kolmogorov-Smirnov testing to conduct comparison on the root-mean-square value and the reference value of the each scanning line in each area, and calculating a corresponding p value; finally using the arithmetic mean p value of the interest area to divide the arithmetic mean p value of the reference value to obtain the corresponding p value. According to the parameter for reflecting biological tissue abnormity and the measuring method thereof, the problems that subjectivity is strong in identification of complex biological tissue abnormity in current B-mode ultrasound diagnosis and rich clinical experiences are needed in the current B-mode ultrasound diagnosis are solved, and measurement is accurate, and implementation is conducted easily.

The invention discloses a mechanical material evaluation method and system. The mechanical material evaluation method and system aim at solving existing technical problems that a traditional scheme is high in subjectivity, and a single decision-making method has limitation. The method includes the following steps that a mechanical material evaluation index system is established, and index weights are determined; the corresponding relationship between evaluation scale language variables and fuzzy measures is determined, material schemes are subjected to expert scoring through the triangular fuzzy number, a fuzzy decision-making matrix is established, the fuzzy decision-making matrix is standardized and weighted, and a weighted standardized judgment matrix is obtained; an ideal solution and a negative ideal solution are determined according to the weighted standardized judgment matrix; the Euclidean distances between the schemes and the ideal solution and between the schemes and the negative ideal solution and grey correlation coefficients are calculated, and the relative close degree and grey correlation similarity of the schemes are calculated according to the result; comprehensive evaluation indexes of the schemes are calculated through a non-linear model according to the relative close degree and the grey correlation similarity in order to carry out priority selection and sorting on the schemes.

The invention discloses a credit scoring method based on consumer group portraits, and relates to the technical field of credit assessment. The method aims to overcome the defect that the existing scoring system is imperfect and the judgment subjectivity is considered to be strong in the prior art. According to the technical scheme of the invention, the method comprises the steps of firstly, classifying acquired consumption data based on consumption data collected by a communication operator and a known credit score corresponding to the collected consumption data; processing the acquired consumption data and the consumption category to which the consumption data belongs through a word2vec model; depicting consumer group portraits, adopting a light GBM as a base learner to learn consumptiondata belonging to the same set, a consumption category to which the consumption data belongs and output information of a word2vec model, training and constructing the credit scoring model; finally, based on the fact that the consumption data belongs to the same set, analyzing the consumption category of the consumption data and the output information of the word2vec model by utilizing the constructed credit scoring model, so that the credit score can be output. The purpose of automatically scoring by utilizing the consumption data is achieved.

The invention provides an image tampering detection method and device based on a dual-channel U-Net model, and aims to solve the problems that in the prior art, a tampering area cannot be effectivelyrecognized, a tampering type cannot be recognized, and a tampering detection technology is single. The U-Net model provided by the invention is an improved U-Net model. A Res-Block structure module, afeature fusion module and a Res-Dilated module are added to the improved U-Net model, so that the use efficiency of features is improved, the information loss is reduced, and a semantic gap between low-dimensional features in an encoder and high-dimensional features in a decoder is solved. According to the method, the defects of singleness and subjectivity of traditional manual feature extractionare overcome, the features extracted through a deep network have higher robustness and better generalization ability, the tampering type of the image can be effectively detected, and the tampering area can be effectively positioned.

The invention belongs to the technical field of computer image processing, and discloses an image definition evaluation method and system, a storage medium, equipment and application. The method comprises the following steps: obtaining a gray matrix of a to-be-evaluated image; calculating the average value of the gray value of a certain pixel point and the gray difference of the adjacent eight pixels, and representing the local definition index of the point by using the product of the average value and the maximum value in the eight gray differences; and calculating the quadratic sum of the definition index of each matrix block to obtain the definition score of the whole picture. According to the method, the problem that the definition evaluation result is inconsistent with the subjectivefeeling of people when a traditional algorithm is used for dealing with certain highly fuzzy pictures can be solved. And compared with a traditional algorithm, the definition of a general picture canbe evaluated more efficiently.

The invention discloses a raw material ratio automatic control method of a cement raw material batching system. A system initial parameter is inputted; the optimal solution is solved by the system according to the initial parameter based on nonlinear programming to calculate the current optimal matching proportion of each raw material, and the obtained latest matching proportion is inputted to a DCS material feeding system again; and when the next period is started, an element online analyzer automatically detects the content of each oxide of the cement raw material mixture, the oxide component of each raw material of the cement is calculated through the least square method, the system automatically updates the exiting oxide component of each raw material of the cement, the optimal solution is solved by using nonlinear programming again to calculate the latest matching proportion and the matching proportion is set in the DCS distributed control system to control a PLC lower computer to feed the material in a periodic cycling way. According to the method, the conventional disadvantages of large data fluctuation and large error caused by manually adjusting the raw material matching proportion of operation personnel by depending on the experience can be avoided so that the stability of the cement raw material components can be effectively guaranteed in real time and the quality of the cement raw material feeding to the kiln can be guaranteed.

The invention relates to an emergency event grading and classification method and device based on a deep convolutional neural network. The method comprises the steps of acquiring image information ofeach category of pre-classified emergency events, and building an emergency event image library; performing importance marking on each image in the image library; selecting a sample set from the marked emergency event image library; building a deep convolutional neural network model, wherein input of the model is an emergency event original image, and output of the model is an emergency event grading and classification vector; inputting the sample set into the model to perform training, and selecting a model by using S-fold cross validation; inputting an image of an emergency event to be graded and classified into the model for calculation, and obtaining a classification and grading result of the emergency event to be graded and classified. According to the method, the deep convolutional neural network is trained through historical image data so as to automatically learn features of an emergency event, thereby performing grading and classification on the event, and assisting a decisionmaker to master conditions of the emergency event scientifically and objectively.

The invention discloses a traffic participant behavior safety assessment method, which can determine the safety level of behaviors according to the basic information of traffic participants and traffic behavior data. The method comprises the following steps of: firstly, establishing a traffic behavior safety assessment index system according to the behavior process of 'pre-event, amid-event and post-event' of the traffic behaviors and according to the principle of two-layer index systems; then determining each safety level of the traffic participants and a standard value of each level assessment factor by comprehensive assessment; designing an assessment index corresponding scale, and acquiring sample data; and finally, based on the research of 'traffic participants, traffic safety assessment indexes and traffic behavior requirements', determining a fuzzy measure model and a fuzzy integral assessment model by using the fuzzy measure theory aiming at an assessment sample structure assessment function, establishing a traffic behavior safety assessment method, and finally obtaining the safety level of the assessed sample by fuzzy identification.

The invention provides a cow weight prediction system, which comprises the following steps of 1, obtaining three-dimensional world coordinates of scenery points through a multi-view stereoscopic vision distance measurement method; 2, calibrating a camera; establishing a relationship between a camera image pixel position and a scene point, obtaining a relationship between the three-dimensional coordinate information of the space object and the two-dimensional information of the acquired image, and further recovering information such as depth and the like missing due to acquisition by utilizingthe relationship; 3, matching feature points; carrying out similarity measurement on the SIFT feature vectors of the two images to be matched, and calculating nearest matching of each feature point ofone image in the feature point set of the other matched image; and 4, determining the body size of the dairy cow according to the obtained feature points, and obtaining a dairy cow weight predictionresult according to a preset prediction model.

The invention discloses a big power grid construction economical type assessment method, comprising steps of establishing an evaluation index system, calculating an index value according to the evaluation index system, wherein the evaluation index system comprises a peak shifting benefit, a mutual aid benefit, a backup benefit, a power supply benefit, an industry benefit, an environment benefit, an operation benefit, a production value benefit and a city benefit, choosing a folding line nondimensionalization mode to perform standardization processing on the evaluation indexes, adopting an addition integration method and a multiplication integration method to perform comprehensive empowerment on various standardized indexes, and establishing a comprehensive model to obtain a comprehensive scores, The invention can guide the electric power enterprise to establish investment benefit philosophy, promotes optimization design scheme and construction scale and is applicable to the big power grid construction economic evaluation research.

The invention provides a cargo weight counting device which comprises an obtaining unit, a recognition unit and a processing unit. The obtaining unit is used for obtaining image data of water gauge scales corresponding to water lines obtained before and after ship unloading, the input end of the recognition unit is connected with the output end of the obtaining unit and used for recognizing water gauge scale values corresponding to waterlines obtained before and after ship unloading from the image data, the input end of the processing unit is connected with the output end of the recognition unit and used for controlling flight motion of the obtaining unit and determining the weight of cargos on a ship according to the water gauge scale values corresponding to the waterlines obtained before and after ship unloading, parameters of the ship and the density of water where the ship is located. The device can control the flight motion speed of the obtaining unit to rapidly and efficiently obtain and calculate needed data, the recognition unit can accurately and scientifically recognize the water gauge scale values corresponding to the waterlines obtained before and after ship unloading, and the problem that subjectivity exists in manual reading is avoided. The invention further discloses a ship cargo weight counting system which has the advantages of the cargo weight counting device.

The invention discloses an image analysis and identification method for lateral flow scrip disease diagnosis; the method comprises the following steps: obtaining an image grey-scale map, and preprocessing same to realize image denoising and luminosity uniformization; using a cross-type template to search maximum or minimum pixel value points, using a threshold method and an edge detection method to search top and bottom boundaries of the detection area according to the searched maximum/minimum value pixel point, and multi-verifying the difference value between the top and bottom boundaries; determining the top and bottom boundaries, setting a certain scope in the left and right sides of the pixel value maximum/minimum value point, using the edge detection method to search the left and right boundaries of a control line, and verifying the width difference between the left and right boundaries; based on the control line position, using a gradient method, a max-min value method and the edge detection method to search the left and right boundaries of a test line in sequence, and multi-determining the reasonability of the searching result. The method can multi-search and verify to-be-identified detection areas, thus improving the analysis and identification precision.

The invention relates to the technical field of crop genetic breeding, particularly relates to a rice variety comprehensive excellent index evaluation method. An entropy weight and efficacy combined analysis method is adopted to screen and evaluate the variety excellence, and the problems of subjectivity, one-sidedness and reliability of a rice variety evaluation method are solved, and the difficulties in comprehensively and objectively analyzing and judging the variety as good, good taste and good processing to meet the requirements of yield, quality, stress resistance and other multi-dimensional characters are overcome. The implementation of the scheme mainly comprises the following steps: (1) establishing an evaluation system; (2) acquiring index data; (3) determining index weights; (4)calculating a comprehensive index; and (5) evaluating excellent varieties.

The present invention discloses a parachuting control motion quantitative evaluation method and system. The method comprises the steps of: establishing a virtual human body model; obtaining attitude data of a head, trunks and limbs of a parachutist in the whole parachuting process; mapping the attitude data to the virtual human body model, and determining motions of a virtual human body in the virtual human body model; and performing comparison and matching of the motions of the virtual human body model and standard motions in an expert database, and giving a quantitative evaluation result according to a matching degree. The method and the system provided by the invention can achieve objective and effective quantitative evaluation of the normativity of the parachuting motions of trainers.

The invention discloses an intelligent terminal online education platform which comprises a user login module, a course reservation module, an identity recognition module, a course evaluation module,a course pushing module, a data storage module, a server, a controller and a camera. The course evaluation module is set, and user scoring serves as the basic score; meanwhile, the obtained sound data, interaction frequency data, courseware page number data and user answer data are used as evaluation influence factors for integrated analysis, the problems that the evaluation mode of teacher scoring is single and subjective influence factors are large are avoided, the fairness of teacher scoring is improved, and the improvement of the teaching quality of teachers is further promoted. The coursepushing module is arranged to count and sort the related subjects browsed and reserved by the user, and the corresponding course information is selected according to the sorting result to be pushed in the pushing time period, so that the deep mining of the education level to the user demand is improved, and the viscosity of the user to the education platform is improved.

The invention discloses an interactive decision-making method based on key historical behaviors in a cloud computing environment. The method includes a basic idea that, before the interaction between a cloud and an entity, both interactive entities respectively send a trust query request to a trust authentication center; the trust authentication center computes the overall trust degrees according to the key historical behavior of the user entity and the service entity, and issues trusts notes to the interactive entities based on the overall trust degrees; both entities perform interactive decision-making according to the trust notes. In the overall trust degree computing process, the trust of the entity is divided into direct trust and indirect trust, and the key historical behaviors are subdivided into a plurality of evaluation indexes by using a granular approach. The method can evaluate trust according to the key historical behaviors of the entities, and the data maintenance and evaluation processes are both executed at the trust authentication center, so that the method is a lightweight class interactive decision-making method. A key historical behavior window is set for trust evaluation, so that the interactive result is more accurate and meanwhile cumulative cheating is also avoided.

The invention discloses an electric power communication network risk assessment system based on clustering and association rule mining, and the method comprises the following steps: collecting historical risk data, building a risk assessment database, and carrying out the preprocessing of the data; performing AP clustering on the database based on the support count of each index, deleting an infrequent index set from the database, obtaining frequent indexes and support counts thereof according to the clustered cluster, and updating the remaining frequent indexes to the database according to adata size sequence; mining an association rule on the updated database, and finding indexes having important influences on an evaluation result; and determining a final index system and the weight ofeach factor by adopting an analytic hierarchy process, and weighting to obtain a final comprehensive risk assessment index. The problems of low mining efficiency of traditional association rules and too strong subjectivity of an analytic hierarchy process are solved, and the accuracy and effectiveness of risk assessment of the power communication network are effectively improved.