118results about How to "Realize accurate identification" patented technology

The invention puts forward a human movement identification method and device, and electronic equipment. The method comprises the following steps that: when a standard movement is displayed, collectingthe video picture frame of the human movement; in the video picture frame, identifying each joint of a human body; connecting two adjacent joints in all joints of the human body to obtain a connecting line between two adjacent connecting lines; calculating a practical include angle between the connecting line between two adjacent connecting lines and a preset reference direction; and according toa difference value between the practical included angle and a standard angle, determining whether the human movement is matched with the standard movement or not. Through the method, the adjacent human joints in the video picture frame of the human body are identified to obtain the connecting line of the adjacent joints, the practical include angle between the connecting line of the adjacent joints and the preset reference direction is calculated, whether the human movement is matched with the standard movement or not is determined according to the difference value between the practical included angle and the standard angle to realize the accurate identification of the movement, and the technical problem in the prior art that movement identification is inaccurate is solved.

The invention relates to a nanopore DNA (Deoxyribose Nucleic Acid) sequencing sensor for detecting DNA deoxyribonucleotide base sequence. The sensor comprises a current detection unit, a molybdenum disulfide field-effect tube, an atomic power microscope feeding system and an array unit, wherein DNA is bonded on an atomic power microscope probe by a chemical modification method; by virtue of a feeding control system of the atomic power microscope, a speed for driving DNA to go in and out of the nanopore is controlled to be 1nanometer per second; the speed fully meets the requirement of bandwidth for detecting DNA sequencing current signal; the nanopore is formed in a molybdenum disulfide nanoribbon in the process of penetrating through the pore of DNA; the molybdenum disulfide nanoribbon plays the role of the field-effect tube in the process of detecting the current signal and can be used for amplifying a signal for showing the DNA penetrating through the hole in real time and effectively increasing a signal to noise ratio; in addition, the method for arraying the atomic power microscope probe corresponding to the molybdenum disulfide field-effect tube can be used for carrying out multi-channel parallel real-time sequencing on the DNA to be detected parallel at the same time, so that the time cost is greatly reduced.

The invention provides an unmanned aerial vehicle data acquisition-based fruit tree identification and quantity monitoring method, which comprises the steps of S1-1, splicing a plurality of images acquired by an unmanned aerial vehicle, and obtaining a three-dimensional point cloud data set and digital surface model DSM data according to the spliced images; S1-2, performing cloth simulation filtering CSF processing on the three-dimensional point cloud data set to obtain digital elevation model DEM data; S1-3, performing abnormal value elimination and DEM calibration on the DEM; and S1-4, calculating a difference value according to the DEM data and the DSM data to obtain tree height data of the orchard, converting the tree height data into a grayscale image, and further converting the grayscale image into a false color image. The invention further provides a corresponding system. The influence of tree shadows and weeds on unmanned aerial vehicle remote sensing image processing is solved, and the fruit tree identification and quantity statistics precision is improved.

The invention discloses a method and device for identifying a store on the basis of user positioning information. The method comprises the steps that a request initiated by a user terminal on the basis of an identifier set in the store is received; the geographic position information of the user terminal is obtained; the geographic position information of the user terminal is obtained according tothe geographic position information of the user terminal, and a store information database is matched to obtain store information; according to the store information obtained through matching, the request is responded to. According to the request initiated by a user on the basis of the identifier set in the store and the geographic position information obtained through positioning of the user terminal, the store information database is matched to obtain the store information, and accurate recognition of the store information can be achieved. The identifiers provided by the store can be a uniform identifier, and people do not need to set different identifiers for different stores, so that the time for setting different identifiers for different stores is also saved.

The invention relates to the field of machine learning, and provides an anomaly recognition method and device based on semi-supervised deep learning, and a storage medium, and the method comprises thesteps: S110, obtaining sample data; s120, acquiring positive sample data enhancement, negative sample data enhancement and data noise; s130, forming a corresponding annotation data positive sample, an annotation data negative sample and an annotation data noise sample; s140, forming three corresponding initial prediction models; s150, respectively inputting the unlabeled sample data into the three trained initial prediction models for data prediction; s160, labeling the unlabeled sample data to form new labeled sample data; s170, adding new labeled sample data into the initial labeled sampledata, and circularly executing the steps S120 to S170 to form a final prediction model; and S180, inputting to-be-identified data into the final prediction model to perform anomaly identification. According to the method, the requirement for data is low, a large amount of marking data is not needed, and meanwhile the data exception recognition accuracy can be improved.

The invention discloses a method and a device for identifying users with abnormal electricity consumption. The method comprises the following steps: acquiring historical power consumption data, and identifying an abnormal value in the historical power consumption data; filtering abnormal values in the historical power consumption data; searching a user electricity consumption characteristic variable in the electricity consumption historical data; and utilizing a preset modeling algorithm to construct an abnormal power consumption user identification model, utilizing the abnormal power consumption user identification modeld to identify abnormal users in the historical power consumption data, wherein input data of the abnormal power consumption user identification model are data corresponding to the user power consumption characteristic variables, and output data are abnormal user identifiers. According to the invention, the technical problems of low detection efficiency and small coverage area of the inspection of the abnormal power utilization users in the related art are solved.

The embodiment of the invention provides a target user terminal recognizing method and device, equipment and a medium. The target user terminal recognizing method comprises the following steps: matching terminal type information of a plurality of user terminals with terminal type information in a target user terminal information set, recognizing a user terminal stealing the type information by using embezzlement feature information, wherein the target user terminal information set includes the terminal type information of the user terminal supporting a target network function after upgrading;and recognizing the target user terminal needing upgrading according to a matching result except the user terminal stealing the type information. Through the embodiment of the invention, wrong recognition of a user can be avoided, e.g., the user terminal supported in default or the user terminal stealing a TAC (Type Allocation Code) is prevented from being wrongly recognized as the user terminal which is supported after upgrading; wrong recognition of the TAC can also be avoided, e.g., the user terminal which cannot be supported after upgrading is prevented from being recognized as the user terminal which is supported after upgrading; and the recognition accuracy is improved.

The invention discloses a one-hole multi-purpose construction method for controlling the impact and gas disaster of deeply buried coal seam. The one-hole multi-purpose construction method includes thefollowing steps that a drilling distance is determined; drilling construction is carried out; a actual stress concentration zone is determined; drilling plugging and gas drainage are carried out; sufficient pressure relief of the coal seam is carried out; and high pressure water injection is carried out. Preferably, the plugging is the first plugging using the expanded capsule, and after the coalseam is fully depressurized, the second complete drilling plugging is performed with sealing agent. According to the one-hole multi-purpose construction method for controlling the impact and gas disaster of deeply buried coal seam, the simultaneous treatment of the deep ground coal seam impact ground pressure and gas disaster can be realized, and the process steps of reasonablely determining thedrilling spacing and the drilling length, the working face drilling pressure relief, the two drilling sealing, the gas drainage, the water injection and the like are adopted, the purpose of multi-purpose, cost reduction and efficiency enhancement are efficiently achieved. The one-hole multi-purpose construction method is simple, easy to operate, can fully utilize the multiple functions of the drilling, and avoids the existing repetitive and wasteful of drilling construction of the existing technology.

The embodiment of the invention discloses a manufacturing method for a display panel. The method comprises the following steps that a substrate is provided, and at least one exposure alignment identification is arranged on the substrate; a colorful filter layer is formed at one side, away from the substrate, of the at least one exposure alignment identification, the colorful filter layer comprisesan identification thickening layer covering the exposure alignment identification, and shapes and sizes of the identification thickening layer and exposure alignment identification are same; a blacklight resistance layer is formed at one side, away from the substrate, of the colorful filter layer; exposure alignment is conducted by utilizing the exposure alignment identification and the identification thickening layer so as to carry out exposure on the black light resistance layer, and a block light shading layer is formed. According to the provided manufacturing method for the display panel, the identification thickening layer covering the exposure alignment identification is formed to increase the height of the position of the exposure alignment identification, and after the subsequentblack light resistance layer is formed, the exposure alignment identification can be accurately identified, therefore the manufacturing accuracy of the black shading layer is improved.

The invention discloses a resume analysis method based on binaryzation. The resume analysis method comprises the following steps: reading resumes in batches; converting the batch read-in resumes intoHTML and TXT text formats; judging whether the resume converted into the HTML format can be applied to a resume accurate identification template or not; according to the method of the regular expression, the resume information converted into the HTML text format is analyzed through the XPATH, the analyzed resume information is scored, and whether the resume information is higher than a predefinedthreshold value or not is judged; using a TensorFlow-constructed BI-LSTM-CRF machine learning model to carry out resume information named entity extraction; extracting and identifying the resume information by utilizing the label dictionary and combining with the named entity, and cutting each block of the resume information; Traversing the content of each resume plate, and storing the extracted resume information content by using a data link table; and storing the resume information subjected to reanalysis by using JSON or XML structured data. According to the invention, the resume information can be accurately identified on the basis of a limited resume sample.

The invention discloses a tail end position compensation method based on robot joint angle compensation and application of the tail end position compensation method. The method comprises the steps that a kinematics model of a robot and an error model of the robot are constructed; an optimal individual is output to serve as an initial solution of a tabu search algorithm by adopting a fitness function; robot kinematics error model identification is conducted by adopting the tabu search algorithm; and robot tail end error compensation is realized by using a Newton-Raphson iteration method. Accurate identification of model errors is realized by using global search capability of a genetic algorithm and local search capability of the tabu search algorithm so as to correct various parameters of arobot model, and Newton-Raphson iteration is applied to compensate for joint angle values of the robot multiple times so that the robot tail end can be positioned to the required position.

The invention discloses a station caption recognition method and a station caption recognition device. The station caption recognition method includes: training a corresponding multilevel classifier according to a pre-generated station caption classification tree with a multilayered structure, wherein the multilevel classifier comprises a plurality of sub-classifiers in one-to-one-correspondence to branch nodes of the station caption classification tree, each sub-classifier comprises a plurality of classification models in one-to-one-correspondence to child nodes corresponding to the branch nodes, positive samples of the classification models are all station caption samples included in the corresponding child nodes, and negative samples of the classification models are all station caption samples included in the other child nodes in brotherhood relation with the corresponding child nodes; recognizing the logo of a to-be-tested station caption in a display image according to the multilevel classifier. The station caption classification tree with the multilayered structure is structured in advance, the corresponding multilevel classifier is generated on the basis of the station caption classification tree, and accurate recognition of the logo of the to-be-tested station caption in the display image can be realized by the aid of the multilevel classifier.

The invention relates to a quadrotor unmanned aerial vehicle parameter distinguishing method based on an improved shuffled frog leaping algorithm. The shuffled frog leaping algorithm is taken as a basis, a frog leaping rule in the algorithm is improved, the global convergence rate of the shuffled frog leaping algorithm is improved, a fitness function is designed according to quadrotor unmanned aerial vehicle parameters, the computation complexity of the algorithm is reduced, and quadrotor unmanned aerial vehicle parameter distinguishing efficiency is improved. The method exhibits certain universality, and parameter distinguishing for the majority of nonlinear systems with a high coupling degree can be realized.

The invention discloses a seed potato processor integrating accurate screening, intelligent identification, dicing and mixing. The seed potato processor comprises a grading feeding system, a dicing system and a mixing system, wherein the grading feeding system divides seed potatoes with different sizes into different grades and conveys the seed potatoes with different grades respectively, and a part of the seed potatoes are conveyed to the dicing system; the dicing system collects information on the number and position of bud eyes on the surface of the seed potatoes based on image recognitiontechnology and dices the seed potatoes in corresponding modes in consideration of nutrient content required for dice growth; the diced seed potatoes are transported to the mixing system, the mixing system performs preservative treatment on the dices to protect the dices. Functions such as grading, feeding, bud eye identification, automatic knife arrangement, dicing, mixing and mixture recycling are integrated, grading of the seed potatoes of different sizes is achieved, and the germination rate of the seed potatoes is ensured; the large seed potatoes are directly separated without dicing, andaccordingly, economic benefits are improved.

The invention relates to a picture recognition method and device, a computer readable storage medium and computer equipment, and the method comprises the steps: carrying out the segmentation of a to-be-recognized picture, and obtaining at least two candidate pictures; extracting a target feature map of each candidate map, wherein the target feature map is a feature map corresponding to target image features of the candidate map; obtaining identification scores of the candidate images under at least two image identification dimensions according to the target feature images of the candidate images; determining a target score of the to-be-identified picture according to the identification scores of each candidate picture under at least two picture identification dimensions; and determining apicture identification result of the to-be-identified picture according to the target score. According to the scheme provided by the invention, the purpose of obtaining the picture recognition resultof the to-be-recognized picture in real time according to the at least two candidate pictures of the to-be-recognized picture is achieved; meanwhile, the plurality of candidate images of the to-be-recognized image are comprehensively considered, so that the recognition accuracy of the image is improved.

The invention relates to an earth and rockfill dam leakage diagnosis method based on Wenner joint inversion, belongs to the field of reservoir dam leakage diagnosis, and also belongs to the field of engineering geophysical detection. The method comprises the steps of: collection of an earth and rockfill dam Wenner joint data volume, joint inversion imaging of the Wenner data volume, correction ofan earth and rockfill dam leakage inversion result and determination of a leakage point according to an image. The method has the advantages that one-sidedness of interpretation and response of different device data bodies to the earth and rockfill dam hidden danger is avoided, accurate identification of the hidden danger body is achieved, and the field work efficiency is improved.

The present invention is applicable to the technical field of dried tangerine peel identification, and provides an identification method and an identification system of dried tangerine peel. The identification method comprises that a sample of dried tangerine peel to be identified is obtained; terahertz spectroscopy information of the dried tangerine peel sample to be identified is obtained by using a terahertz spectrometer; and according to the terahertz spectroscopy information of the dried tangerine peel sample to be identified, the variety and storage period of the dried tangerine peel sample to be identified are determined by using the terahertz spectroscopy difference information of pre-obtained dried tangerine peel with different varieties and storage periods. The accurate identification of dried tangerine peel variety and storage period is realized, and the technical problems that the obtained information of the dried tangerine peel sample is single and the variety and storageperiod of dried tangerine peel cannot be accurately determined are solved.

The invention discloses a green channel vehicle cargo carrying radioactive source image recognition method based on a convolutional neural network. The method comprises the following steps: step 1, preprocessing a green channel vehicle cargo carrying radioactive source image; 2, preparing an input image sample; step 3, designing a green channel vehicle cargo carrying radioactive source image recognition model; step 4, optimizing the green channel vehicle cargo carrying radioactive source image identification model; and step 5, by training, verifying and testing the model and recording changesof a loss function and classification accuracy in the training process, the loss function can reflect the capacity of the model for accurately classifying the cargo types. According to the invention,the green channel vehicle carried cargo radioactive source image identification based on the convolutional neural network is adopted, so that subjective dependence of an inspection result on inspection personnel can be avoided, and the working intensity of front-line inspection personnel is reduced. And meanwhile, the inspection efficiency can be improved, and the congestion condition of the tollstation is reduced.

The invention discloses a traffic signboard detection method based on deep learning, and belongs to the technical field of computer vision image processing. The method is characterized in that the method comprises the following steps: a, collecting pictures of traffic signboards on a road under various weather and illumination conditions, and carrying out target detection and classified label labeling; b, performing data preprocessing on the image through an image processing module; c, after data preprocessing, adopting a YOLOv3 model in target detection as a detection network, and sequentially performing model building, model training and model tuning to complete training; d, inputting a to-be-detected picture into the trained model to obtain a prediction result of the signboard position and the classification label in the current picture. According to the method, the YOLOv3 model is used as a detection network, model building, model training and model optimization are carried out in sequence, accurate identification of the traffic sign in an all-weather scene can be realized, and the method has good generalization ability.

The invention discloses an intelligent analysis method for identifying unreasonable power-failure management. The intelligent analysis method comprises the steps of (1) establishing a unified full-path power grid equipment library, and taking as unique identification of equipment in the whole network; (2) establishing a power grid full-topological model; (3) establishing an equipment power-failure history library; (4) extracting power-failure equipment and establishing a power-failure analysis model; (5) analyzing a power-failure influence range; (6) identifying whether the equipment suffers from an overlapped power-failure operation or not; (7) identifying a repeated power-failure operation; and (8) forming a power-failure plan analysis report, and forming a power-failure analysis report from the aspects of the power-failure influence range, whether the equipment suffers from the overlapped power-failure or not and whether the equipment suffers from the repeated power-failure or not. By adoption of the intelligent analysis method for identifying the unreasonable power-failure management, through intelligent analysis on a power grid graphical model, the reasonability of power-failure plan working arrangement is accurately formed from transverse and longitudinal dimensions, so that error and omission caused by human subjective judgment are avoided, and accurate identification of a power-failure plan risk evaluation operation is realized.

The invention relates to a method for identifying Baijiu of different years through ultra-high performance liquid chromatography-tandem mass spectrometry. The method comprises the following steps of:filtering Baijiu to be identified, respectively measuring in ultra-high performance liquid chromatography quadrupole electrostatic field orbital ion trap mass spectrometry, collecting data, carrying out background drift correction, peak extraction and peak alignment, and identifying compounds; performing partial least squares discriminant analysis on a result data set, screening the data by takinga variable importance value greater than 1 as a standard to obtain characteristic compounds of Baijiu of different years to be identified, taking the characteristic compounds as an input layer of theartificial neural network, and taking the Baijiu of different years to be identified as an output layer of the artificial neural network; and sequentially performing artificial neural network training, verification and testing to obtain an artificial neural network identification model, finally acquiring characteristic compounds with variable importance values greater than 1 in the Baijiu to be identified, and inputting the characteristic compounds into the artificial neural network identification model to obtain an output layer which is the year of the Baijiu to be identified.

The invention relates to a bus load identification method and system. The method comprises the steps of collecting load identification data of a to-be-identified bus; determining load power time sequence data corresponding to various electrical loads on the to-be-identified bus according to the load identification data of the to-be-identified bus. According to the technical scheme provided by theinvention, the identification of the load types of large-scale and diversified loads on the bus is realized through the learning capability of the LSTM neural network, and the identification precisionis improved.

The invention discloses a deep learning-based ship board character accurate detection method, and the method comprises the steps of obtaining a target image in real time, inputting the target image into a ship detection classification model, and obtaining a ship image containing ship board characters, wherein the ship detection and classification model detects and classifies an input target imageto obtain a first image containing ship board characters, positions a ship in the first image to obtain a positioning position of the ship in the first image, and intercepts the first image accordingto the positioning position to obtain a ship image containing the ship board characters; preprocessing the ship image to obtain a preprocessed image; inputting the preprocessed image into a ship boardcharacter recognition model to obtain ship board character information, wherein the shipboard character recognition model is used for performing character text recognition on an input preprocessed image and outputting shipboard character information. According to the technical scheme of the invention, the method achieves the precise recognition of the small ship side characters, and improves therecognition accuracy and detection efficiency of the ship side characters.

The invention discloses an intelligent operation method of a server. A voice recognition function is added to the server. Through speech recognition, an operator is associated with a user name in a system. When a control instruction input by a user through voice is obtained, firstly, user identity recognition is conducted according to a voice control instruction. After a user currently operating aserver is obtained, a server control strategy under the user is determined from pre-configured user server control strategies, and user operation is processed according to the strategies. Operationsof the server are identified and processed on the basis of the current operation user, so that the real-time operator is accurately identified, the security threat to the data of the server caused bythe operation of other users after the login of a certain user identity is avoided, and service quality of the server is improved. The invention further discloses an intelligent operation device of the server, an intelligent server and a readable storage medium which have the above beneficial effects.

The invention relates to a text recognition method, a computer readable storage medium and computer equipment. The method comprises the steps of extracting at least two words in a to-be-recognized text; obtaining a word code of each word under a semantic dimension and word codes of each word under at least two text classification dimensions; performing connection processing on the word code of each word under the semantic dimension and the word code of each word under at least two text classification dimensions to obtain a target matrix; determining the classification probability of the to-be-recognized text under each text classification dimension according to the target matrix; and determining a text recognition result of the to-be-recognized text according to the classification probability of the to-be-recognized text under each text classification dimension. According to the scheme provided by the invention, semantic information and interpretability of word codes of the text are comprehensively considered, and the text is discriminated from multiple text recognition dimensions, so that text recognition is more accurate, and the text recognition accuracy is improved.

The invention belongs to the field of the Internet and communication, and particularly relates to a harassing call intercepting method based on a combination of the Internet technology and the communication technology. The method comprises the following steps: (1) when a calling number comes in, a calling signaling message of the calling number is transmitted to an interception system, the interception system judges whether the number is a harassing number, if yes, the second step is carried out, and if not, the third step is carried out; (2) the interception system returns an EndCall message for call interruption, and the fourth step is carried out; (3) the interception system returns a Continue message for call continuity, and the fifth step is carried out; (4) the information of intercepting the calling number is sent to a terminal of a called user, and the fifth step is carried out; and (5) the interception process is over. The method of the invention realizes the problem that through how to effectively combine the Internet technology and the communication technology, recognition and interception of harassing calls at the cloud end can be realized accurately and in a real time mode, and the user can be announced in an effective Internet means.

The invention discloses a matrix, a preparation method thereof and a biological sample testing method and belongs to the field of mass spectrometry. The matrix can be used for matrix-assisted laser desorption ionization mass spectrometry. Aggregate of a particulate matter is contained in the matrix and has monodispersity, the particulate matter is of a spherical core-shell structure with a coarsesurface, and particle size is smaller than 1um. The particulate matter comprises a core layer, a first shell layer and a second shell layer coated by the first shell layer, the core layer is ferroferric oxide, the first shell layer is silicon dioxide, and the second shell layer is a silver-platinum compound. The matrix is easy to obtain and highly resistant to salt, and can simplify the biologicalsample pretreatment process and improve detection on desorption ionization of micromolecules when being used for mass spectrometry, so that high sensitivity can be realized within a wide detection range, and biological samples needed for consumption are few.

The invention discloses an intelligent safe fitness guidance device, system and method. The system comprises an interactive display device, a camera device, an instrument motion state acquisition device, a server and an action recognition processor. The interactive display device comprises a main control board and a display; the main control board is respectively connected with the display, the camera device, the instrument motion state acquisition device, the server and the action recognition processor; the action recognition processor is used for recognizing user actions, and the interactive display device is used for receiving user touch operation and displaying recognized action data and corrected action data in real time; the server comprises a network communication module and an action correction module, the action correction module is connected with the network communication module, and the action correction module is used for performing action comparison according to the standard motion model. According to the invention, the human body model is established to accurately record the training postures of people in real time, and the standard posture database is matched to timely feed back the user to correct the actions of the user.

The invention discloses an event detection method and a related device, and the method comprises the steps of inputting an image into an instance segmentation network model, carrying out the analysisof the image through the instance segmentation network model, and outputting feature information; calculating a first area intersection ratio between segmentation instance masks in the feature information corresponding to the pedestrian; comparing the first area intersection ratio with a first preset threshold value, and if the first area intersection ratio is larger than the first preset threshold value, considering that a trailing event exists; calculating a second area intersection ratio between the segmentation instance mask in the feature information corresponding to the pedestrian and the segmentation instance mask in the feature information corresponding to the target area; calculating the average value of the second area cross-parallel ratio in the preset time; and comparing the average value with a second preset threshold value, and if the average value is greater than the second preset threshold value, determining that a staying event exists. The method can accurately and effectively identify the trailing event and the staying event.

The invention relates to an MEMS gyroscope parameter identification driving control method based on data screening, and belongs to the field of intelligent instruments and meters. The method comprisesthe following steps: converting a gyroscope dynamic model into a dimensionless dynamic linear parameterized model; designing an online data screening method to select proper historical data, and constructing a parameter adaptive law in combination with the current data; using an interval excitation mode to loosen strict requirements of parameter identification on continuous excitation conditions;and designing a controller to realize parameter identification and driving control at the same time. According to the MEMS gyroscope parameter identification driving control method based on data screening, the problems that true values are difficult to obtain in parameter identification and the requirements for continuous excitation are strict can be solved, the accurate dynamic model is obtained, meanwhile, high-precision gyroscope driving control is achieved, and the performance of the MEMS gyroscope is further improved.