94results about How to "Solve the inaccurate identification" patented technology

The invention provides a method for network equipment identity recognition. The method for the network equipment identity recognition includes the following steps: receiving visit requests sent by network equipment through a webpage browser; searching corresponding webpage codes and collection codes according to an address of a target webpage and returning the codes to the webpage browser; receiving an attribute vector set collected from the network equipment and sent by the collection codes through the request of hyper text transport protocol (HTTP); searching attribute groups matched with the attribute vector set from a preset attribute bank based on the attribute vector set; carrying out similarity computation on the attribute vector set and the searched attribute groups and selecting the attribute group which has high similarity with the attribute vector set; and comparing the highest similarity value with a preset new equipment threshold value and an old equipment threshold value and determining the type of the network equipment according to comparison results. The invention further provides a network equipment identity recognition system for realizing the method. The method and the system for network equipment identity recognition can reduce occupation on network equipment resources, and improve visit speed and recognition accuracy.

The invention discloses a method for identifying irregular wave troughs of a pulse wave signal, belonging to the technical field of pulse wave signal identification. The method comprises the following steps of: identifying one section of input pulse signal and extracting one typical pulse wave period; intercepting data of a declining branch part in the typical period, and carrying out vertical coordinate transformation to convert the traditional stepped untypical wave crests and wave troughs into typical wave crests and wave troughs and further identify the positions of maximum values and minimum values of the transformed data; and finally, amending the positions of the identified maximum values according to data which are not transformed, and marking the positions of the finally-identified wave crests and wave troughs of pulse wave and outputting the positions for analysis. According to the method disclosed by the invention, the missed identification and mistaken identification of a pulse condition analytical instrument on untypical wave crests and wave troughs can be solved, the stepped untypical wave crests and wave troughs can be effectively identified and new hardware equipment is prevented from being modified or added; and the method is particularly suitable for the traditional pulse condition analysis instrument.

The embodiment of the invention discloses a power equipment image recognition method and device based on transfer learning and a neural network. The method comprises the steps of acquiring a simulation image data set of target power equipment to serve as a first training sample, the simulation image data set comprises simulation images of the target power equipment and corresponding classificationlabels, wherein the number of models of each type of target power equipment is multiple; obtaining an actual image data set of the target power equipment as a second training sample, wherein the actual image data set comprises an actual image of the target power equipment and a corresponding classification label; training a pre-constructed convolutional neural network model by using the first training sample to obtain a first target convolutional neural network model; training the first target convolutional neural network model by using the second training sample to obtain a second target convolutional neural network model; and inputting the image of the to-be-identified power equipment into the second target convolutional neural network model for identification. And the accuracy of powerequipment image identification is improved.

Techniques for performing audio-visual speech recognition, with improved recognition performance, in a degraded visual environment. For example, in one aspect of the invention, a technique for use in accordance with an audio-visual speech recognition system for improving a recognition performance thereof includes the steps / operations of: (i) selecting between an acoustic-only data model and an acoustic-visual data model based on a condition associated with a visual environment; and (ii) decoding at least a portion of an input spoken utterance using the selected data model. Advantageously, during periods of degraded visual conditions, the audio-visual speech recognition system is able to decode (recognize) input speech data using audio-only data, thus avoiding recognition inaccuracies that may result from performing speech recognition based on acoustic-visual data models and degraded visual data.

The invention relates to a labeling quality detection method used for carrying out qualified detection on an automatically-pasted label. The labeling quality detection method comprises following steps of carrying out image acquisition on the automatically-pasted label; carrying out initial processing on the acquired image; carrying out presence detection on the processed image, judging whether the label exists or not, entering the next step when the label exists, and determining the label to be unqualified when the label does not exist; carrying out character detection on the processed image, judging whether the character in the image is consistent with the specified character of the label, indicating the label to be qualified when the character in the image is consistent with the specified character of the label, and indicating the label to be unqualified when the character in the image is not consistent with the specified character of the label. Compared with the prior art, the labeling quality detection method is advantaged by being convenient for detection, accurate in examination, low in cost and convenient to implement and the like.

The embodiment of the invention discloses a traffic indicator light state identification method, device, an electronic device and a storage medium. The method comprises the following steps of: receiving the information of the traffic indicator sent by the radio frequency tag arranged on the traffic indicator; the state recognition of the traffic light is performed in an image currently captured bythe camera according to the information. The technical scheme avoids the problem of inaccurate identification of traffic indicator lights or inaccurate identification of traffic indicator lights dueto inaccurate vehicle positioning information or delayed updating of high-precision maps, and further improves vehicle driving safety.

Disclosed is a first-order sea clutter detection method based on least squares approximation. The invention relates to a first-order sea clutter detection method, and aims to solve the problem that the found Bragg peak position is inaccurate and the false alarm rate is high in the prior art. The method comprises the following steps: calculating the theoretical Bragg frequency fB based on the working frequency f0 of radar, defining a noise zone on each sea element, and calculating the threshold of each sea element; detecting sea area sea state data according to the radar, determining the maximum radial velocity Vm of ocean current, and calculating the maximum Bragg frequency offset Delta fm; determining a search scope according to Delta fm and fB, determining a suspected point of the Bragg peak, confirming the threshold of the amplitude of the suspected point, and judging whether the current sea element is effective; for the effective sea elements, taking the neighboring points of an appropriate number of suspected points to form a point set, and acquiring the approximation function of the point set using a least squares approximation method; and finally, taking the peak point of the approximation function as a Bragg peak. The first-order sea clutter detection method is applied to the field of high-frequency ground wave radar detection.

The invention relates to the technical field of artificial intelligence, and provides an identity card picture processing method and device, a terminal and a storage medium, and the method comprises the steps: cascading a semantic segmentation network and a classification network to obtain a target network, and training a multi-task model based on the target network; inputting a picture containingan identity card into the multi-task model; obtaining an identity card mask picture and a category identifier output by the multi-task model; calculating coordinates of four vertexes of the identitycard in the identity card mask picture; and according to the coordinates of the four vertexes and the category identifier, converting the identity card mask picture into a standard identity card picture in a perspective manner. According to the method, the identity card with the forward character orientation can be accurately segmented. In addition, the invention also relates to a blockchain technology, and the standard identity card picture can be stored in a blockchain node.

The invention discloses a method, device and equipment for calibrating lane line information based on a radar and a storage medium. The method comprises the steps of obtaining static target information at the two sides of a road through radar equipment installed on a vehicle; clustering the static target information, and carrying out curve fitting on the classified clusters; and performing feature matching on the fitted curve and lane line information recognized by the camera, if matching succeeds, executing a normal vehicle control instruction, and if matching fails, calibrating the lane line information according to the fitted curve. According to the method for calibrating the lane line information provided by the embodiment of the invention, the millimeter wave radar is adopted to extract the road information, mutual verification with the lane line information identified by the camera can be carried out under the condition of not increasing a system sensor, and the lane line identification accuracy is greatly improved.

The invention relates to a sign-in method based on multi-mode recognition. The sign-in method comprises the steps of: acquiring face video images and fingerprint information of identity information recognizing people; automatically detecting infrared rays on the surface of human bodies by means of a pyroelectric infrared sensor which is arranged on the surface of sign-in equipment, and controlling the sign-in equipment to turn on by the pyroelectric infrared sensor when staff enters a sign-in sensing region; selecting the face video image under the appropriate illumination intensity background according to the illumination intensity during sign in, matching the selected face video image with acquired facial image data, and matching fingerprints of the staff with acquired fingerprint information, and calculating a matched-degree score result A; and selecting the face video image under the appropriate illumination intensity background again according to the illumination intensity during sign in, matching the selected face video image with the acquired facial image data, and matching the fingerprints of the staff with the acquired fingerprint information, and calculating a matched-degree score result B. The sign-in method has the advantages of accurate judgment, good reliability, low price and wide application.

The invention is suitable for the field of terminal artificial intelligence and the technical field of corresponding computer vision, and provides a language recognition method and a device, terminalequipment and a computer readable storage medium. The method comprises the steps that a to-be-recognized text line image is acquired, and the to-be-recognized text line image comprises a to-be-recognized text; inputting the text line image to be recognized into a trained language recognition model, obtaining the language of the text to be recognized; wherein the language identification model is used for determining the language of the to-be-identified text according to the to-be-identified text line image, and if the language is a language family language comprising multiple languages, takingthe language corresponding to the language rule in the language family language as the language of the to-be-identified text according to the language rule of the to-be-identified text. The language is recognized based on the language rule, the problem of inaccurate recognition caused by ambiguity of the same or similar characters is avoided, and the accuracy of language recognition is improved.

The invention provides an identification method and device. The identification method comprises the steps: obtaining a to-be-identified character image; inputting the to-be-recognized character imageinto a character recognition model, and extracting character features of the to-be-recognized character image through a feature extraction layer in the character recognition model; generating an intermediate feature vector according to the character features, and processing the intermediate feature vector through a full connection layer in the character recognition model to obtain a multi-dimensional feature vector; inputting the multi-dimensional feature vector into an output layer in the character recognition model for vector conversion, and outputting a recognition result of characters in the to-be-recognized character image. According to the recognition method provided by the invention, the correctness of the handwritten characters can be detected while the handwritten characters are recognized, so more application scenes are enriched, and the recognition efficiency and the recognition accuracy can be effectively improved.

The invention discloses an instruction identification method and electronic equipment. The instruction identification method is applied to the electronic equipment; the electronic equipment comprises an information acquiring unit and a keyboard; the instruction identification method comprises the steps that at least one part of the keyboard is divided into at least two sub zones, the sub zones are encoded, and the keyboard division result and the coding of each sub zone are stored; the information acquiring unit captures the coverage situation of each sub zone; element values are assigned for the sub zones according to the coverage situations of the sub zones; an input instruction is determined according to the coding of each sub zone and the corresponding element values of the sub zones. With the adoption of the instruction identification method, the input instruction can be obtained through accurate and efficient hand gesture recognition, and the input efficiency is further improved.

The invention discloses an action recognition method and device, and relates to the technical field of computers. The method comprises the following steps: acquiring a video clip, and determining at least two target objects in the video clip; for each target object in the at least two target objects, connecting the positions of the target object in each video frame of the video clip, and constructing a space-time diagram of the target object; dividing at least two space-time diagrams constructed for at least two target objects into a plurality of space-time diagram subsets, and determining a final selection subset from the plurality of space-time diagram subsets; and determining the action category between the target objects indicated by the relationship between the space-time diagrams contained in the final subset as the action category of the action contained in the video clip. By adopting the method, the accuracy of action recognition can be improved.

The invention discloses a license plate intelligent recognition method, which comprises the following steps: S1 obtains a vehicle image containing a license plate, and preprocesses the image; S2, roughly locating the license plate region of the preprocessed image; S3, accurately locating the license plate of the image after the rough localization of the license plate area; S4, carry out characterclipping process on that license plate image obtained by fine positioning; S5, normalize that cut character image and extracting feature; S6 adopts template matching algorithm to automatically recognize license plate characters. The invention solves the problem of inaccurate positioning of the inclined license plate, and has the advantages of high recognition accuracy and good recognition effect on the inclined license plate.

The invention provides a method and device for establishing a feature library based on gait recognition and face recognition, and the method comprises the steps of firstly obtaining a monitoring video, and screening out a monitoring image of each pedestrian feature from the monitoring video, secondly, extracting the face features of the pedestrians, and determining the occurrence time of the monitoring images of the pedestrian features in the monitoring video, determining a gait feature video which contains the monitoring image in the monitoring video and has a video duration of a preset timeperiod through the occurrence time of the monitoring image in the monitoring video and a preset time period, and extracting the gait features of pedestrians in the gait feature video, and finally, establishing the feature library according to the face features of each pedestrian and the gait features corresponding to the face features of each pedestrian, so that the accuracy during the identity recognition process is improved.

The embodiments of the invention disclose a vertical serial number identification method and a vertical serial number identification device. The method comprises the following steps: getting a vertical serial number area image under ultraviolet fluorescence; performing clustering according to the gray values of the pixels in the vertical serial number area image; correcting the gray values of the pixels according to the clustering result to increase the gray difference between the pixels; determining the position of a vertical serial number target image in the corrected vertical serial number area image; and identifying each vertical serial number character in the vertical serial number target image. By implementing the embodiments of the invention, the problem that vertical serial number identification based on the existing vertical serial number identification method is inaccurate is solved, and the accuracy of vertical serial number identification is improved.

The invention provides a handwritten numeral recognition method and a handwritten numeral recognition device. The method comprises the following steps of: receiving a handwritten numeral image input by a user; extracting M covariance characteristics of the handwritten numeral image, wherein the value of M is any one positive integer which is greater than 1; and recognizing handwritten numerals according to the preset M covariance characteristics of each training image, a preset type tag carried by each training image and the M covariance characteristics of the handwritten numeral image. The handwritten numerals are recognized by extracting the M covariance characteristics of the handwritten numeral image, wherein the value of the M is any one positive integer which is greater than 1, so that the problem of inaccurate recognition of the handwritten numerals caused by adoption of a single covariance characteristic in the recognition process of the handwritten numerals in the prior art is solved.

The invention relates to a method and a device for judging the connectivity of sand bodies between injection and production wells of a low-permeability reservoir. The method comprises the steps of: fitting to obtain a starting pressure gradient permeability curve according to a starting pressure gradient corresponding to the core test permeability of a target region; taking a permeability value corresponding to the maximum variation amplitude of the starting pressure gradient along with the variation of the permeability on the curve as the limit water injection permeability, and taking the permeability value corresponding to the maximum curvature on the curve as the free-flow water injection permeability; and filtering a sandstone section interpreted by the target region by adopting the permeability greater than or equal to the ultimate water injection permeability and less than or equal to the free-flowing water injection permeability to obtain a longitudinal sand body connection diagram and a planar sand body distribution diagram between the injection and production wells, and judging the connectivity between the injection and production wells in the research region. According tothe method and the device, a rock core starting pressure gradient experiment is associated with the dynamic connectivity of the sand bodies between the oil-water wells of the oil reservoir, and the originally explained sandstone section is subjected to secondary filtering explanation by adopting the permeability within a certain range, so that the connectivity between a water injection well and each oil production well is quantitatively judged.

The invention relates to a voice recognition method and device, electronic equipment and a storage medium. The voice recognition method comprises the following steps: acquiring an original voice signal; performing noise reduction on the original voice signal to obtain an enhanced voice signal; respectively extracting voice features of the original voice signal to obtain a first voice feature, andextracting voice features of the enhanced voice signal to obtain a second voice feature; jointly processing the first voice feature and the second voice feature by using a pre-trained acoustic model to obtain a combined state sequence; and decoding the combined state sequence to obtain a voice recognition result. The original voice signal and the voice signal after noise reduction are processed together, so that the accuracy of voice recognition is improved.

The invention provides a pet image emotion recognition method based on a deep residual network, and the method comprises the steps: obtaining the audio and video of a single pet under different emotions, carrying out the preprocessing according to the audio, and extracting the sound features; performing statistical classification and emotion marking on the audio according to the sound features, and extracting a corresponding single-frame picture according to a time sequence and the video to obtain a pet image emotion data set with emotion marks; constructing a pet emotion classification algorithm model, and taking the pet image emotion data set as training data to perform model training; and performing target detection and emotion deduction on the pet through the trained model. According to the method, the determinacy of pet emotion recognition can be improved, and the inference speed and precision of recognition are improved.

The invention relates to the technical field of robots, and specifically relates to a routing inspection robot control method, a routing inspection system, a storage medium and an electronic device. The method comprises the following steps: acquiring an routing inspection image; identifying a to-be-inspected target in the routing inspection image; when the to-be-inspected target in the routing inspection image is identified out, judging whether the routing inspection image is matched with a preset image, if the routing inspection image is matched with the preset image, confirming that the to-be-inspected target is normal; and if the routing inspection image is not matched with the preset image, confirming that the to-be-inspected target is abnormal. Thus, inspection for abnormal conditionsis realized, and routing inspection efficiency and accuracy are improved.

The present disclosure provides an image processing method, comprising: acquiring an image sequence, the image sequence comprising a plurality of images, each image of the plurality of images comprising m*n pixel points, m being a positive integer, and n being a positive integer; for pixel points at specific positions in the m*n pixel points, establishing k models for the pixel points at the specific positions according to the first i images in the image sequence, wherein i is a positive integer, and k is a positive integer; and performing matching processing on the pixel point at the corresponding position of at least one image in the image sequence according to the k models of the pixel points at the specific positions, and determining whether the pixel point at the corresponding position is a moving target pixel point or not according to a matching result.

The embodiment of the invention discloses an information processing method and device for hip joint CT image recognition, and the method comprises the steps: firstly carrying out the preprocessing of an original hip joint CT image, inputting the preprocessed image into a 2D DenseUnet neural network model, carrying out the recognition of a fracture region, inputting the image data of the recognized fracture region into a MaskRCNN network model, segmenting a fracture region by the MaskRCNN to generate image data of a fracture block region; and finally, mapping the image of the fracture area to the original image scale of the CT image. An original hip joint CT image is recognized and segmented through a 2D DenseUnet neural network model and a MaskRCNN network model, and accurate recognition of a fracture block is achieved.

The invention discloses a parameter identification method for a grid-connected photovoltaic power generation system based on transient disturbance. The grid-connected photovoltaic power generation system comprises a photovoltaic array, a DC / DC converter, an inverter controlled by a VSC, an inverter, an inverter controller, and a filter circuit. The filter circuit is connected with a power grid. The DC / DC converter is connected with a maximum power controller. The inverter is connected with the inverter controller. The method is based on an improved genetic algorithm-particle swarm optimization(GA-PSO). By analyzing current changes under different transient disturbances, optimal identification of inverter parameters is achieved. The problem that a traditional mathematical analysis method is inaccurate in identification of the inverter parameters is solved, and a foundation is provided for research on inverter control optimization, low voltage ride-through, power grid power flow optimization and the like.

The invention discloses a meridian acupoint identification method. The method comprises the following steps: acquiring a human body image of a target user; determining a target position of a target meridian acupoint in the human body image in the standard human body meridian diagram according to the human body image and the standard human body meridian diagram; and based on the target position, identifying the actual position of the target meridian acupoint on the human body. By applying the technical scheme provided by the invention, through combination of the human body image and the standard human body meridian diagram, the meridian acupoints on the human body can be accurately identified, the problem of inaccurate meridian acupoint identification caused by individual difference is effectively solved, and a better health management effect is favorably achieved. The invention further discloses a meridian acupoint identification device and equipment and a storage medium, and the corresponding technical effects are achieved.

The invention discloses an accurate inclination character recognition method based on a robustness template. The method comprises the following steps: establishing a sample image data set; establishing a plurality of robust character templates corresponding to the character categories; performing multi-scale construction on the to-be-recognized inclination character image block and each robustness character template; calculating the similarity between the to-be-identified dip angle character image block under each scale and each template character image block; determining an initial character category of the to-be-recognized dip angle character image block according to the similarity value under each scale; filtering the inclination character image block to be identified and the robustness character template of the initial character category thereof by using a convolution kernel; calculating and filtering the similarity of the to-be-recognized dip angle character image block and the filtering template character image block; and determining the character type of the to-be-recognized inclination character image block. The invention discloses an accurate inclination character recognition system based on the robustness template. According to the invention, the accuracy of inclination character recognition is effectively improved.

The invention relates to the technical field of artificial intelligence, and discloses a character recognition method and device based on a real plate, computer equipment and a storage medium, and themethod comprises the steps: starting a preset infrared light generator and a visible light generator; performing image acquisition processing on the real plate to obtain a first image; if the color value contrast of the second character pattern in the first image is greater than a preset contrast degree threshold, obtaining a second text; turning off the infrared light generator, and turning on apreset ultraviolet light generator; performing image acquisition processing on the real plate to obtain a second image; if the color value contrast of the first character pattern in the second imageis greater than a contrast degree threshold, obtaining a first text; and combining the first text and the second text to obtain a final character recognition result. Therefore, the accuracy of character recognition is improved. In addition, the invention also relates to a blockchain technology, and the character recognition model can be stored in the blockchain.

The invention relates to a method for determining sweeping direction of a sweeper and a device thereof and the sweeper. The method comprises the following steps: acquiring point cloud data in a target range through a three-dimensional space sensor, wherein the point cloud data comprises multiple pieces of distance information between the sweeper and an obstacle; extracting a plane in a target range based on the point cloud data; and taking a target plane meeting a preset condition in the planes as a wall surface, and determining a normal direction of the wall surface as a main sweeping direction of the sweeper. The point cloud data of the surrounding environment of the sweeper is collected through the three-dimensional space sensor, the multiple planes in the three-dimensional space are constructed and extracted through the point cloud data, the wall surface is recognized from the multiple planes in the three-dimensional space, then the normal direction of the wall surface serves as the main sweeping direction of the sweeper, and the problems that the wall surface recognition is inaccurate, and the main sweeping direction of the sweeper is not accurately found can be solved.