334results about How to "Improve recognition efficiency" patented technology

The present invention relates to an electric power device identification model construction method and system, and an identification method of an electric power device. The electric power device identification model construction method comprises: marking electric power device targets in an infrared image respectively corresponding to each type of an electric power device, and obtaining a sample training set; inputting the sample training set into a RPN convolutional neural network to allow a loss function to have a minimum value, and outputting a target candidate frame; inputting the target candidate frame into a Fast-RCNN convolutional neural network, calculating the conversion weight value of the target candidate frame to a corresponding type according to a fully connected layer and a regression function, and employing the frame regression to obtain the Fast-RCNN parameters of the position of the target candidate frame being offset to a corresponding tag position; and setting a sharing convolutional layer learning rage as 0, and performing initialization of the RPN convolutional neural network and the Fast-RCNN convolutional neural network, performing training of an input infrared image in the RPN convolutional neural network according to the Fast-RCNN parameters, and obtaining a RPN convolutional neural network model; and inputting the target candidate frame into the RPN convolutional neural network model, updating the Fast-RCNN convolutional neural network to form a uniform Fast-RCNN network, and outputting an electric power identification model.

The invention relates to a multi-meter character recognition system and a multi-meter character recognition method based on machine vision. The system comprises an image acquisition module, a network storage module, and a software processing module. The software processing module comprises an electric meter group segmentation unit, an image preprocessing unit, an image tilt correction unit, an image enhancement unit, a character area positioning unit, a character segmentation unit, an image normalization unit, a character feature extraction unit, and a BP neural network identification unit. The image acquisition module comprises a grayscale processing unit, a binary processing unit, and a morphological processing unit. Multiple electric meter images can be identified at the same time, and human resource and work time are saved greatly. As an FTP cloud is used to store electric meter images and electric meter identification data, resource sharing is saved, the problem on storage of a lot of images and a lot of electric meter data collected in real time by cameras in a production field is solved, the memory of PC is released, and the efficiency of electric meter identification is increased.

The invention discloses a method and system for naming a picture and belongs to the field of computer technologies. The method comprises the first step of obtaining the picture, source information of the obtained picture and content information of the picture and finding one or more preset pictures, corresponding to the content information of the picture, in a feature library according to the content information of the picture, the second step of performing image matching on the obtained picture and one or more found preset pictures corresponding to the content information of the picture and selecting one similar picture most similar to the obtained picture, the third step of obtaining feature information corresponding to the similar picture, wherein the feather information is used for describing one or more features of the obtained picture, and the fourth step of naming the obtained picture. Through the source information of the picture and the content information of the picture, the matching range of the feature information of the picture in the feature library is determined, the picture is named according to the matched feature information, the source information of the picture and the content information of the picture, and therefore the recognition efficiency of a user for searching the pictures is improved.

The invention provides an unmanned aerial vehicle identification system and an identification method thereof. The identification system comprises a flight controller (1) which is arranged on an unmanned aerial vehicle and an identification terminal (2) which wirelessly communicates with the flight controller (1). The flight controller (1) comprises an information transmitting module (3). The identification terminal (2) comprises a wireless communication module (4) and a processing module (5). The processing module (5) transmits an identification request through the wireless communication module (4). The information transmitting module (3) responds to the identification request and transmits an identification code uniquely corresponding to the unmanned aerial vehicle to the identification terminal (2). The processing module (5) transmits a marking task request to the unmanned aerial vehicle corresponding to the identification code. The flight controller (1) performs a marking task according to the marking task request.

The invention relates to a ring tone recognition method and system for a call center system. The method comprises the steps of firstly, calling the input telephone number; then, initiating a fax request to a called user, and preliminarily recognizing a fax machine via SDP (Session Description Protocol) analysis; recognizing a ring tone state, wherein based on an audio fragment within 5 seconds after calling, the ring state before the called user answers being standard ring-back tone or color ring tone is recognized; and after the called user hooks off, recognizing fax machine off-hook, automatic responder off-hook, natural person off-hook and get-through nobody answer based on voice analysis of a voice fragment after off-hook answer. By adopting the method, the number state and the terminal type can be accurately recognized, the calculation quantity of ring tone state recognition is reduced, the recognition result is quickly given, and the recognition efficiency is improved.

The embodiment of the invention discloses a program recognition method and device based on machine learning. The method comprises analyzing an inputted unknown program and extracting class behavior features of the unknown program, the class behavior features including library feature and application programming interface API (Application Program Interface) feature of an import table; coarsely classifying the unknown program according to the extracted class behavior features; inputting the unknown program to a generated training model and a corresponding decision machine to judge the unknown program according to the coarse classification result; and outputting the recognition result which shows that the unknown program is a malicious program or a non-malicious program. Based on machine learning technology, the method provided by the invention can obtain a model for recognizing malicious programs based on class behaviors by extracting and analyzing class behavior features of a large amount of program samples, and the model can save a large amount of man power and can improve malicious program recognition efficiency.

The embodiment of the invention discloses a vehicle recognition method and device and a storage medium According to the embodiment of the invention, the method comprises the steps: carrying out the combination of each two collected vehicle sample images, so as to build sample pairs; merging each sample pair into one multichannel image, and then adding the multichannel images to a training sample set; training a preset twin neural network model according to the training sample set, and obtaining a trained model, so as to achieve the recognition of a to-be-recognized image based on the trained model during the vehicle recognition, such as the recognition of a target vehicle from mass images, and so on. According to the scheme of the invention, the method can achieve the recognition of a vehicle through the building of a model. Compared with a conventional scheme based on human eyes or simple matching, the method can improve the recognition efficiency and accuracy.

The invention provides a vehicle speed measuring method and a system based on RFID (Radio Frequency Identification). The method comprises that: antennas of at least three RFID systems emit radio frequency signals outwardly with same emission time and emission angles, and the at least three RFID systems receive response signals of the radio frequency signals which carry vehicle information and are sent by a running vehicle; and the at least three RFID systems forward the response signals to a speed calculating system which calculates running speed of the vehicle in regions between any two of the three RFID systems according to time of the response signals sent by the at least three RFID systems and distance between the at least three RFID systems and the speed calculating system. Speed of multiple vehicles running in the same direction can be measured simultaneously, and speed measuring regions of the vehicles can be flexibly confirmed according to geographical positions of the RFID systems so that identification efficiency is high and installation and use cost is low.

A hardware trojan identification method based on a single classification supporting vector machine belongs to the detection and identification field of hardware trojan chips. The invention solves the problems of huge error and low efficiency in the prior art because the technology of utilizing chip side channel information to identify the chip hardware trojan requires to observe an image manually. The method provided by the invention comprises the following steps: I, preprocessing, and acquiring a side channel information matrix; II, selecting a backward analysis chip to perform backward analysis, and determining whether a hardware trojan is contained; III, dividing the backward analysis chip not containing the hardware trojan into a train sample and a train optimizing sample; IV, utilizing the side channel information matrix of the train sample to establish the characteristic space of the chip; V, acquiring the side channel characteristic data matrix of a chip to be detected; VI, performing normalization processing; VII, picking up the normalized data of the train sample and the train optimizing sample; VIII, training the single classification supporting vector machine to constitute a minimum hypersphere; and IX, if beyond the minimum hypersphere, then determining that the chip to be detected is a hardware trojan chip.

A method of improving multi-type barcode recognition efficiency using continuous sampling is applicable to a barcode recognition apparatus. According to the method, each barcode type is given with a corresponding weight value, and a total weight threshold is set in the barcode recognition apparatus. A barcode pattern is sampled with a limited number of times, and each sampled image sample is only compared with a part of the supported barcode types. When the comparison matches, a weight value is assigned to the matched barcode type, and the weight value is accumulated to acquire a total weight value of the matched barcode type. When comparison of all image samples is finished, the barcode type with the total weight value reaching the total weight threshold is the barcode type of the barcode pattern, and a recognition result of the barcode pattern is output according to the barcode type.

The invention relates to a GIS partial discharge on-line monitoring mode identification method and a GIS partial discharge on-line monitoring mode identification system. The method comprises the following specific steps: processing received partial discharge data to obtain the amplitude, number of discharges and phase of the partial discharge data; generating a PRPS three-dimension spectrogram according to the number of discharges, amplitude and phase, and further generating a PRPD spectrogram, a v-phi spectrogram, and an n-phi spectrogram; extracting the characteristic value of the partial discharge data based on the PRPD spectrogram, the v-phi spectrogram and the n-phi spectrogram; and carrying out person correlation calculation on the characteristic value of the partial discharge data and the characteristic value of each partial discharge data type in a sample library one by one to identify the type of partial discharge. The GIS partial discharge on-line monitoring mode identification system of the invention comprises an acquisition unit and a data processing unit. The method and the system of the invention have the advantages of simple identification process and high identification rate, and have great value of on-site engineering application.

The invention discloses a method and a system for partitioned input of a handwritten character string, belonging to the field of handwriting input. The method is used on user equipment comprising a man-machine interaction panel. The method comprises the following step: step 10, collecting chirography from the handwriting area of the man-machine interaction panel; step 20, identifying whether the collected chirography is a character string chirography or a command gesture chirography; step 30, if the collected chirography is the character string chirography, identifying the collected chirography in a character string identification range corresponding to the character string input area in which the chirography exists; and if the collected chirography is the command gesture chirography, determining the command corresponding to the chirography, and executing corresponding operation. Due to the method and the system for realizing the method, the identification efficiency of handwriting is improved and the input efficiency is also improved.

The invention provides a medical image abnormal region segmentation method and device. The method comprises the steps of obtaining a medical image; dividing the medical image into a plurality of to-be-classified image blocks, and utilizing a first machine learning model to classify each to-be-classified image block, the classification result being a normal image block or an abnormal image block; respectively taking to-be-segmented image blocks from the medical image based on the abnormal image blocks, and respectively performing semantic segmentation on each to-be-segmented image block by using a second machine learning model, the segmentation result being an abnormal pixel point; and marking the abnormal pixel points in the medical image.

The invention discloses a method and an apparatus for searching for a target person in a video. The method for searching for the target person in the video comprises the steps of receiving target person information and a to-be-checked video, wherein the target person information comprises a name and/or a photo of the target person; determining a template corresponding to the target person according to the target person information; segmenting the to-be-checked video into a plurality of key frame images; screening out a plurality of human face-containing frame images from the key frame images based on a human face identification algorithm; and determining a target person-containing frame image in the human face-containing frame images according to the template, and obtaining time information, in the to-be-checked video, of the target person-containing frame image. According to the method and the apparatus for searching for the target person in the video, provided by embodiments of the invention, a sensitive person in the video can be quickly and conveniently found out, so that the identification efficiency is improved.

The invention discloses an automatic identifying and classifying method for sensitive data, comprising the steps of: pre-treating a training dataset, removing unmeaning words, and thereby obtaining a word set; treating the word set according to TF-IDF, calculating the vector weighed value of each word, and thereby obtaining a basic data corpus; manually identifying and classifying the basic data corpus, selecting the words on which sensitive data can be marked, and thereby forming a sensitive word corpus; performing word segmentation on target data by the word segmentation technology, and then matching extracted features with the sensitive word corpus; and sorting the classification according to accumulated values of the weighted values of the sensitive words from high to low. According to the automatic identifying and classifying method for sensitive data, through establishing the sensitive word corpus, performing word segmentation on the target data, and calculating the accumulated values of the weighted values of the sensitive words, the classification of the sensitive data is achieved, and thereby efficiency and accuracy of the sensitive data identification and classification are improved, and the defect that the existing sensitive data is identified depending on dictionary classification and manual classification is remedied.

The invention discloses a target graphic code identification method and device. The method comprises a step of: if a target image of a graphic code identification area comprises at least two graphic codes, adding a triggerable control to any of the at least two graphic codes, so as to identify a target graphic code in the at least two graphic codes in the target image through responding a triggeroperation for the triggerable control corresponding to the target graphic code in the at least two graphic codes.

The invention discloses a point cloud extraction method and device for a columnar object. The method includes the steps of: merging collected single frames of point clouds into a dense point cloud set; searching a line type point cloud block in a preset direction from the dense point cloud set; slicing the line type point cloud block to obtain piece point clouds; searching from the piece point clouds at least one candidate piece point cloud whose periphery is empty space; determining at least one columnar object point cloud according to the at least one candidate piece point cloud; and identifying the at least one columnar point cloud according to a preset columnar object model. The method analyzes the piece point clouds to obtain the adhesion condition of each piece point cloud of the columnar object with the other object, thereby obtaining the adhesion condition of the point clouds of the columnar object with the other object, and achieving an effect of identifying columnar object point clouds adhered to the other object. An effect of identifying from road point clouds a columnar object adhered to the other object is realized, and the identification efficiency of the columnar object is improved.

The invention provides an access control method and device, and applied to network equipment. The method comprises the following steps of: analyzing message characteristics of a service message after receiving the service message sent by client equipment; judging whether the message characteristics are matched with a protocol identification table or not, wherein the protocol identification table comprises corresponding relationships between the message characteristics and protocol types; determining the protocol type of the service message according to a matched table item if the message characteristics are matched with the protocol identification table; and performing access control processing of the service message according to the protocol type. Due to application of the embodiment of the method and device, after the network equipment receives the service message sent by the client equipment, the protocol type of the service message can be rapidly identified according to the protocol identification table based on the message characteristics of the service message; the service message is unnecessary to be subjected to tedious operations, such as reading of a load content and application layer information reorganization; therefore, consumption of a CPU of the network equipment is low; and the identification efficiency is high.

The present application relates to an article identification method, apparatus, computer readable storage medium and computer device. The method includes: acquiring at least one image of a target article in an article placement space; inputting an image of a target object into an image detection model; acquiring image feature distribution information outputted from the image detection model, wherein the image feature distribution information comprises image features of articles corresponding to each target article and feature position information corresponding to the image features of each article in the image of the target article; matching the image features of each item with the standard item features in the item database to obtain the corresponding matching degree; confirming a targetarticle type corresponding to each article image characteristic according to the matching degree; according to the image feature distribution information and the target article type, the article position and the corresponding article type of each target article in the article placement space being identified. The method can not only improve the recognition efficiency, but also improve the recognition accuracy.

The invention discloses an architectural-drawing-based automatic identification method for a column and a wall. Information of a layer at which each column is located and a layer at which each wall islocated is read, coordinate data of end points of column lines of all columns are found out according to a pixel of each expression column, and the coordinate data are written into a column point setcorrespondingly; according to all processed new wall lines, whether all closed polygonal wall lines are connected with columns is determined; if not, coordinate data of end points of the closed polygonal wall lines are found out and are written into wall point sets correspondingly; if so, coordinate data of end points of a wall-column pixel formed by connection of the closed polygonal wall linesand the columns are written into wall-column point sets correspondingly; and a corresponding column contour line is generated based on each column point set and the generated line is outputted, a wallcontour line is generated based on each wall point set and the generated line is outputted, and a wall-column contour line is generated based on each wall-column point set and the generated line is outputted. Therefore, the recognition efficiency and recognition accuracy of the column and wall of the building object can be improved.

The invention discloses a gang member account recognition method, device and equipment, and the method comprises the steps: obtaining an initial suspected member account of a target business, and selecting a target member dimension relation table for the target business; generating correlation between target suspected member accounts according to the initial suspected member account and the target member dimension relation table; recognizing target gang member accounts according to the correlation between target suspected member accounts, wherein one suspected member account and at least another suspected member account in target gang member account have the direct relation. According to the embodiment of the invention, the process of gang member account recognition can be automatically completed through recognition equipment, so the recognition efficiency is relatively high. Moreover, because the method can recognize a latent suspected member account in the gang member accounts, so the method improves the accuracy of a recognition result.

The invention aims to provide an ares protocol analysis system based on a peer-to-peer network, and a realization method thereof. The problem that now P2P flow becomes the burden of the network can be solved by deploying a P2P cache server, and a protocol identification system in a cache server architecture prevents the problems of high power consumption, instability, low identification efficiency and the like of the traditional protocol analysis system. According to hash searching as well as the size and the message characteristic word of an application layer message of some interactive messages, the message is identified with high efficiency, then the {source address, download resource number} information of a hash searching request message is analyzed and extracted and sent to an indexserver, simultaneously the corresponding hash searching request message is copied and cached in a queue so as to wait the returned node list information of the index server, and finally a hash searching return result message is constructed to an ares client of the peer-to-peer network according to the information returned by the index server.

The invention discloses a temperature response type europium ion imprinting composite membrane preparation method and belongs to the technical field of environmental functional material preparation. Specifically, a regenerated cellulose membrane is utilized as a supporting membrane, graphene oxide and modified silicon dioxide nanosphere are utilized as membrane preparing materials, an europium ion is utilized as a formwork, acrylic amide is utilized as a functional monomer, N-isopropyl acrylamide is utilized as a temperature sensitive type functional monomer, ethylene glycol dimenthacrylate is utilized as a cross-linking agent, azodiisobutyronitrile is utilized as an initiator, a composite membrane is prepared on the surface of the supporting membrane, a two-step temperature polymerization method is utilized to prepare a temperature response type europium ion imprinting composite membrane. A static adsorption experiment and a selective permeation experiment are utilized to research adsorption balance, adsorption kinetics, selective recognition and europium ion separation performance of the prepared temperature response type europium ion imprinting composite membrane. Results show that the temperature response type europium ion imprinting composite membrane prepared through the method disclosed by the invention has higher specific adsorption capacity on europium ions and excellent europium ion recognition and separation performance under condition of 35 DEG C.

Provided are a method, device, and mobile terminal for identifying malicious advertisements in a mobile terminal, and a storage medium. The method comprises the following steps: obtaining an application program running in a mobile terminal; monitoring a notification bar of the mobile terminal to determine notification information corresponding to the application program; and, according to the frequency of appearance of the notification information and/or the content of the notification information, determining whether the application program displays malicious advertisements. The present method determines the notification information corresponding to an application program by monitoring the notification bar of the mobile terminal, and determines whether the application program displays malicious advertisements according to the frequency of appearance of the notification information and/or the content of the notification information, thereby accurately identifying application programs capable of displaying malicious advertisements, effectively compensating for the deficiencies of determining malicious advertisements by extraction code feature matching technology, and increasing the efficiency of malicious advertisement identification.

The invention relates to a multichannel-based method for preventing reader collision in an RFID system. The method comprises the following steps of: analyzing the collision types of readers, and constructing a solution for various reader collision problems by combining the information of distances among readers; in order to better execute the solution, establishing a criterion of optimizing data channels aiming at different relationships among the readers; setting a message format of control information so as to enable related readers to conveniently obtain selection information of the channels; in order to better complete communication between the readers and labels, designing a novel communication processing flow so as to better perform a multichannel function; and furthermore simulating and verifying the performance advantages of the method in the aspects of recognizing the command failure number, recognizing the processing time and network load and the like in comparison with existing agreements. The method provided by the invention is suitable for RFID actual scenes with a plurality of readers, and plays a role in pushing the information acquisition technology.

The invention provides a website automated testing method and an automated testing system to solve the problem that the automated testing efficiency is low at present. The website automated testing method comprises the steps of analyzing web pages, generating middleware variables corresponding to page elements in the web pages, generating middleware interfaces according to the middleware variables, receiving to-be-tested page element information, searching for the middleware interface matched with the to-be-tested page element information, and calling the matched middleware interface to test the to-be-tested page element. The identifying efficiency of the page element is improved, so that the automated testing efficiency is improved.

The invention provides a method for constructing a DNA sequencing library of a genome under detection, comprising the steps: (1), subjecting the genome under detection to first digestion by using micrococcal nuclease to obtain a first digestion product; (2), subjecting the first digestion product to co-immunoprecipitation to obtain a co-immunoprecipitation product; (3), subjecting the co-immunoprecipitation product to second digestion by using protease K to obtain second digestion product; (4), subjecting the second digestion product directly to dephosphorylation by using shrimp alkaline phosphatase to obtain a dephosphorylation product; (5), subjecting the dephosphorylation product directly to denaturalization to obtain a denaturalization product with single-branched DNA molecules; and (6), acquiring the sequencing library by TELP (tailing, extension, ligation and PCR) method based on the denaturalization product with single-branched DNA molecules.

The invention provides a mobile terminal. The mobile terminal comprises a shell, a support part, a flexible display screen and a fingerprint identification module, wherein the shell has a top surface and side surfaces, which are connected; a mounting groove is formed in the side surface and used for mounting the fingerprint identification module; one end of the support part is connected with the top surface, and the other end of the support part extends out of the top surface; the flexible display screen comprises a fixed part and a moving part which are connected; the fixed part clings to the top surface; the moving part clings to the support part; and when the flexible display screen is folded, the support part is bent in a preset shape, so that the fingerprint identification module performs fingerprint identification through the support part and the flexible display screen. According to the mobile terminal, the identification efficiency and accuracy of the fingerprint identification module are improved.

The embodiment of the invention provides a network traffic identification method and apparatus, and relates to the technical field of the Internet. The method comprises the following steps: identifying cloud service traffic in the received network traffic; determining traffic that satisfies a bridge keep-alive mechanism in the cloud service traffic to serve as the target traffic; extracting a feature value of the target traffic for a traffic feature in a traffic feature set; inputting the extracted feature value into a pre-trained anonymous network traffic identification model to identify whether the target traffic is anonymous network traffic, wherein the anonymous network traffic identification model is a random forest trained according to a part of traffics in a sample network traffic set and a part of traffics in the traffic feature set, the random forest comprises at least one decision tree, one node of the decision tree corresponds to one traffic feature, and the traffic featurecorresponding to the node has a value range. By application of the network traffic identification method provided by the embodiment of the invention, the workload in a network traffic identification process can be reduced, and the identification efficiency can be improved.

The invention relates to a character information identification method for a medical image and a device thereof. The method comprises the steps that input of selecting a character area of an operator on the medical image is received, and the character area under character information identification is confirmed according to the input; image preprocessing operation is performed on the character area so that non-character information is removed and a clean character image is obtained; the character image is cut so that characters are split out and single character images are obtained, and the obtained single character images are identified so that preliminarily identified characters are obtained; the preliminarily identified characters are corrected; and the corrected identified characters are verified according to the preset verification conditions so that unreasonable and illegal identified characters are removed, a verification result is outputted, and the verification result includes the verified identified characters. The identification range is reduced according to the received input of the operator, identification efficiency is enhanced, and the preprocessed and preliminarily identified characters are corrected and verified so that identification accuracy is further enhanced.