416results about How to "Solve technical problems with low accuracy" patented technology

The invention discloses a recommendation method and device of media content, a storage medium, and an electronic device. The method comprises the following steps: acquiring a recommendation request, wherein the recommendation request is used for requesting to recommend media content to the target object; acquiring a subject word from the word of the first content in response to the recommendationrequest, wherein the first similarity of the word vector of the subject word and the semantic vector of the first content is greater than or equal to the second similarity, the semantic vector of thefirst content is determined according to the word vector of each keyword in multiple keywords and the weight of each keyword, multiple keywords are the keywords in the words of the first content; selecting the second content matched with the subject word from candidate media content, and recommending the second content to the target object. The technical problem that the recommended media contentis low in accuracy rate in the related technology is solved.

The invention provides a speech feature extracting method and a speech feature extracting device base d on artificial intelligence, wherein the method comprises the following steps: conducting spectral analysis on to-be-recognized speech, and conducting feature extraction on obtained spectrogram of the to-be-recognized speech by virtue of a VGG (Vancouver global green) convolutional neural network, so that speech features are obtained. By virtue of the spectrogram, the to-be-recognized speech can be described in an image mode and the VGG convolutional neural network is an effective way for image processing, so that the speech features, which are extracted on the basis of the mode, can be used for accurately describing speech characteristics. The technical problem in the prior art that speech recognition is poor in accuracy since the speech features cannot be effectively extracted when the speech recognition is conducted in accordance with the extracted speech features.

The invention discloses a data processing method and a data processing device. The data processing method comprises the following steps: acquiring a code file which is a source program text including a character string; performing lexical analysis on the character string in the code file, thereby acquiring a lexical unit sequence; analyzing the code file, thereby acquiring a preset object; relating the lexical unit sequence to the preset object, thereby establishing an overall symbol table which is used for recording the data information of all the preset objects in the code file; performing static code scanning on the code file according to the overall symbol table, thereby acquiring a scanning result, wherein the scanning result at least comprises a search result for the category of the lexical unit sequence. The invention solves the technical problem of low accuracy of the code scanning in the related technique.

The invention discloses a motion track reconstruction method and device, a storage medium and an electronic device. The method includes the following steps: obtaining a data set, wherein positioning data obtained by positioning a target object during movement of the target object is stored in the data set; fitting a part of the positioning data in the data set to obtain multiple sections of firstcurves, wherein the first curve are obtained by fitting a plurality of positioning data with adjacent acquisition time and a proportion of noise data in the other part of the positioning data of the data set is higher than a proportion of noise data in the part of the positioning data of the data set; and determining a motion track of the target object based on the multiple sections of the first curves. The motion track reconstruction method and device, the storage medium and the electronic device solve the technical problem of low accuracy rate of reconstruction of the motion rack of a vehicle in the related art.

The invention provides a voice characteristic extracting method and apparatus based on artificial intelligence. The method includes the following steps: analyzing frequency spectrum of to-be-identified voice, after obtaining a spectrogram of the to-be-identified voice, using an out-gate convolutional neural network to extract characteristics of the spectrogram, and obtaining voice characteristics. Since the spectrogram can describe the to-be-identified voice by using images, and out-gate convolutional neural network is an efficient way to process images, the voice characteristics extracted by the method herein can describe the characteristics of voice in an accurate manner. The method can address in accuracy of voice identification caused by the lack of effective extraction of the voice characteristics in the prior art.

The invention discloses a video scene recognition method and device, a storage medium and an electronic device. The video scene recognition method comprises the steps: obtaining a target video to be subjected to scene recognition; performing frame extraction sampling on the video sequence of the target video to obtain a sub-video sequence, and performing frame extraction sampling on the audio sequence of the target video to obtain a sub-audio sequence; extracting a target video feature from the sub-video sequence, and extracting a target audio feature from the sub-audio sequence; processing the target video features through the first target model to obtain a first classification result, and processing the target audio features through the second target model to obtain a second classification result; and determining a target scene of the target video based on the first classification result and the second classification result. According to the invention, the technical problem of low recognition accuracy of the video scene in the prior art is solved.

The invention discloses a method and device for determining a point of interest, a storage medium and an electronic device. The method includes the following steps: obtaining a first positioning request, wherein the first positioning request is used to request to acquire a target point of interest where a target account is located; searching for candidate points of interest corresponding to location information of the target account in response to the first positioning request; and selecting the target point of interest matching spatial context features and temporal context features of the target account from the candidate points of interest, wherein the spatial context features are used to represent a spatial correlation relationship between the points of interest, and the temporal context features are used to represent a temporal correlation relationship between the points of interest. According to the scheme of the invention, the technical problem that the accuracy of positioning the point of interest (POI) of the user is relatively low in related technologies can be solved.

The invention discloses a text processing method and device and a storage medium. The method comprises the steps of obtaining a plurality of original short sentences of an original text and a plurality of original words of each original short sentence, and obtaining a plurality of target short sentences of a target text and a plurality of target words of each target short sentence; determining anoriginal short sentence vector of each original short sentence according to the original word vector of each original word and the corresponding importance weight, and determining a target short sentence vector of each target short sentence according to the target word vector of each target word and the corresponding importance weight; obtaining the similarity between the original text and the target text according to a first target matrix determined by each original short sentence vector and the corresponding importance weight and a second target matrix determined by each target short sentence vector and the corresponding importance weight; and when the similarity is greater than a target threshold, determining that the target text is matched with the original text, and outputting the target text, so that the technical problem that the text matching accuracy is lower in the prior art is solved.

The invention discloses a method and device for determining an abnormal interactive account. The method comprises the steps that interactive data among a plurality of accounts is obtained; based on the interactive data, an interactive network of the accounts and at least one key account of the interactive network are determined; at least one interactive group containing the key account in the interactive network is determined; based on feature information of the interactive group, it is determined whether or not the interactive group is an abnormal interactive group, and the abnormal interactive group contains an abnormal interactive account. The method and device for determining the abnormal interactive account have the advantages that the technical problem in the prior art is solved thataccount containing abnormal interaction is determined just by the amount of the interactive content among accounts so that the accuracy of the result is lower.

The invention discloses a gesture recognizing method, device and terminal. According to an embodiment, the method includes responding to the gesture operation of a user, and acquiring the trace of the gesture operation; matching the trace of the gesture operation with each preset gesture template to acquire the corresponded degree of similarity; judging whether the acquired highest degree of similarity is not smaller than a preset threshold or not; if so, determining the gesture template with the highest degree of similarity to be a relevant gesture template of the gesture operation to execute the operation corresponding to the relevant gesture template. By the aid of the method for recognizing the user's gesture operation, the technical problems that in the prior art, the recognizable gestures are limited and the gesture recognizing rate is low are solved, and the using efficiency of the terminal is improved.

The embodiment of the invention discloses a terminal log grasping method, device and system, solving the current technical problem that the information amount of operation logs is great, a cloud receives the logs passively, therefore the terminal log grasping accuracy is low; the terminal log grasping method of the embodiment of the invention comprises: periodically scanning a database, determining there is an untreated strategy record in the database; determining the corresponding client according to the untreated strategy record, building connection relation with the client, sending the log information corresponding to the strategy record and requested to be reported to the client; and obtaining the appointed log which is reported by the client according to the log information requested to be reported and corresponds to the log information requested to be reported.

The invention discloses an LED display screen splicing error detection method and a detection device, wherein the method comprises the following steps: acquiring an image of a spliced region of an LED display screen; recognizing first LED pixel points and second LED pixel points from the image; acquiring the location of the first LED pixel points in the image and the location of the second LED pixel points in the image; and on the basis of the location of the first LED pixel points and the location of the second LED pixel points, determining a splicing error of the LED display screen, wherein the image is obtained by shooting the spliced region; the LED display screen comprises a first display unit and a second display unit; the spliced region is obtained by splicing the first display unit and the second display unit; the first display unit comprises a plurality of the first LED pixel points; and the second display unit comprises a plurality of the second LED pixel points. The detection method and the detection device disclosed by the invention can solve the technical problem that an LED screen, acquired by related technologies, is relatively low in splicing error precision.

The invention discloses an abnormality identification method, device and system for a biological organ tissue image. The method comprises the steps of obtaining a first image frame, wherein the firstimage frame is an image which is obtained by shooting a living body and comprises organ tissues of the living body; dividing the first image frame into a plurality of first areas through a target model, searching a target area from a candidate area taking a point in the first area as a center in the first image frame; wherein the target area is a candidate area where the abnormal part of the organtissue of the organism is located in the first image frame, and the target model is a pre-trained neural network model used for identifying the area where the abnormal part of the organ tissue is located in the image frame; and identifying a target area in the first image frame. According to the present invention, the technical problem that in the prior art, the accuracy of abnormal part detection is low is solved.

The invention discloses a model training method and device and a recognition method and device of a statement service type. The model training method comprises the steps of collecting multiple piece of text information of each service type in multiple service types, and marking the service type which each statement in the collected text information belongs to so as to obtain a first training set; determining a statement vector corresponding to each statement in the first training set, wherein the statement vector is a multi-dimensional array used for solely representing the corresponding statement; utilizing a support vector machine to train a statement vector corresponding to the statement of each service type in the multiple service types to obtain a training model of the support vector machine corresponding to each service type, wherein the training model of the support vector machine is used for recognizing the service type of the statement. According to the model training method and device and the recognition method and device of the statement service type, the technical problem in the prior art that the service type recognition of the statement is low in accuracy is solved.

The invention discloses an address identification method and device and a storage medium. The method comprises the steps of obtaining an address fragment sequence obtained after segmentation of a to-be-identified address text; obtaining a splicing feature vector matched with the address fragment sequence; inputting the splicing feature vector into an address identification network model, and obtaining an output result output by the address identification network model, wherein the address identification network model is obtained by deeply learning a plurality of training samples; according tothe output result, identifying an address role label sequence matched with the address fragment sequence, wherein the address role label sequence includes address role labels matched with all addressfragments included in the address fragment sequence respectively. The technical problem that in the related technologies, the address identification result is low in accuracy is solved.

The invention discloses a distribution data prediction method and device and a terminal. The method comprises the steps of obtaining order information; According to the order information and the pre-estimation model, obtaining pre-estimated delivery time; Selecting a corresponding deliveryman according to the order information, and generating a delivery path of the deliveryman; And generating a delivery scheme according to the predicted delivery time, the deliveryman and the delivery path. The technical problem of low accuracy caused by large deviation between data received by the background in take-out software and actual data in the prior art is solved.

The invention discloses a method and a device for identifying an object in an image based on artificial intelligence and a cloud server, a storage medium and an electronic device. The method comprisesthe steps that after a first picture and a second picture are obtained, a first feature map of the first picture and a second feature map of the second picture are obtained, the first picture comprises a first object, and the second picture comprises a second object; obtaining each pixel in the first feature map and each pixel in the second feature map to obtain a first pixel set; determining thesimilarity between any two pixels in the first pixel set to obtain a two-dimensional matrix; adjusting the similarity smaller than a first preset threshold value in the two-dimensional matrix into the first preset threshold value to obtain a two-dimensional adjacent matrix; and processing the two-dimensional adjacency matrix to obtain a first identification result. According to the invention, thetechnical problem of low accuracy of comparing whether the objects in the picture are the same object is solved.

The invention discloses a processing method and device for guideboard information. The processing method comprises the steps that laser point data contained in laser point cloud data of a target road is clustered according to density to obtain at least one laser point data set; a periphery convex hull of the laser point data in the sets is acquired; and the guideboard information on the target road is recognized based on the shape and vertex angle of the periphery convex hull. Through the processing method and device, the technical problem that a guideboard detection method is low in precision is solved, and the guideboard information can be acquired accurately and efficiently.

The embodiment of the invention discloses a target user recommendation method, and the method comprises the steps: extracting a first sample set and a second sample set, enabling the first sample setto comprise the features of a user group in a source field, and enabling the second sample set to comprise the features of the user group in a target field, wherein the data size of the features in the source domain is greater than that of the features in the target domain; utilizing the first sample set and the second sample set to train a basic neural network model in a transfer learning mode toobtain a target neural network model; extracting features of a to-be-recommended user and taking the features as input of the target neural network model, and then obtaining a recommendation probability of the to-be-recommended user; and enabling the target neural network model to sort the recommendation probabilities of the to-be-recommended users, determine the target users according to a sorting result, and complete recommendation of the target users.

The embodiment of the invention discloses a voice information processing method, which is applied to a terminal. The method comprises steps; voice signals are acquired; the voice signals are recognized, and corresponding text information is acquired; and based on a session history database preset in the terminal, correction is carried out on the test information, and the text information after correction is acquired, wherein the session history database stores session history records of the user. The embodiment of the invention also provides a terminal.

The invention discloses a fire monitoring method, a fire monitoring system and a fire monitoring server. The method comprises the following steps: reading a thermal image of a target area monitored at the current moment and collected monitoring data; calculating out a suspicious fire threshold according to the grey value of the thermal image and the monitoring data; when there is any one or more pixels of which the gray values are greater than the fire threshold in the thermal image, determining that there is a suspicious fire area in the target area; reading a thermal image of the target area monitored after the current moment; and extracting the dynamic parameters of the suspicious fire area in the thermal image of the target area monitored at the current moment or in the thermal image of the target area monitored after the current moment, and acquiring fire information of the target area after judging whether the dynamic parameters meet a first preset condition. The technical problem in the prior art that the accuracy of fire monitoring is low due to the influence of external conditions in the process of fire monitoring is solved.

The invention discloses a road scene map building method and device. The method comprises the steps that an editing instruction editing vector map data is received, wherein the editing instruction isused for indicating that target operation is executed on an operation object in the vector map data; the operation object is subjected to target operation in a two-dimensional scene or a three-dimensional scene according to the editing instruction, and an operation result is obtained, wherein coordinates in the two-dimensional scene or the three-dimensional scene has the mapping relation; information of the operation object in the vector map data is updated according to the operation result, updated information of the operation object in the vector map data is obtained; according to the updated information, the information of the operation object in a target scene is updated, wherein the target scene is the scene not executing target operation on the operation object in the three-dimensional scene and the two-dimensional scene. Accordingly, the technical problem that the built road scene map is low in precision is solved.

The invention discloses a method and device for detecting brake shoe earring faults of railroad trains. The method for detecting brake shoe earring faults of railroad trains comprises the following steps of: obtaining a wheel axis coordinate of a railroad train from an image of the railroad train; obtaining a segment related to a brake shoe earring from the image of the railroad train; and constructing a brake shoe earring positioning and recognition model by adoption of a method combining deep learning and a support vector machine; and detecting the segment related to the brake shoe earring according to the brake shoe earring positioning and recognition model so as to detect brake shoe earring faults of the railroad train. According to the method and device disclosed by the invention, the technical problem that the existing method is not high in brake shoe earring fault detection correctness is solved, and the technical effect of improving the brake shoe earring fault detection correctness is realized.

The invention belongs to the field of electronic technologies, and relates to a digital pulse width modulator. The digital pulse width modulator comprises a differential clock selection module, a coursing regulating module, a fine modulating module and a PWM signal generation module, wherein the coursing regulating module determines frequency for outputting PWM waves according to a synchronous counter, the differential clock selection module selects an input clock of the fine regulating module, the fine modulating module uses a ripper counter to generate an end position control signal of the PWM signals, and the PWM signal generation module controls output of the PWM signals according to the control signals of the coursing regulating and fine regulating modules. Therefore, the technical problem that in the prior art, PWM accuracy is low is solved, and the technical effects of improving the PWM accuracy and reducing a dependence degree of generation of the high accuracy PWM waves to a programmable logic gate chip are achieved.

The invention provides an image recognition method and device based on an image recognition model. The method comprises the steps that according to combination of different running environments and food material placement strategies of a refrigerator, food materials in the refrigerator are imaged, and multiple food material images are obtained; every time before imaging, refrigerator doors are opened and closed at least once; according to the imaged food materials in each food material image, the food material images are annotated; training samples are generated according to multiple annotatedfood material images; the training samples are adopted to train the image recognition model; and the trained image recognition model is adopted to perform image recognition. Due to the fact that thefood material images are obtained through imaging of the food materials in the refrigerator according to the combination of different running environments and the food material placement strategies ofthe refrigerator, when the annotated food material images are utilized to generate the training samples to train the image recognition model, the accuracy of food material recognition through the image recognition model can be improved, and the problem that accuracy is low when an existing intelligent refrigerator recognizes food is solved.

The invention discloses a train control method and device, a storage medium and a processor. The method comprises the steps of obtaining the jumping distance and the jumping direction of a train; determining rotation information of a veering frame axle of the train according to the jumping distance and the jumping direction; controlling the train to move to a target position to stop according to the rotation information. By means of the train control method and device, the storage medium and the processor, the technical problem that the accuracy of control over train jumping is low is solved,and therefore the accuracy of the control over train jumping is improved.

The invention discloses an inquiry method and an inquiry device for fault causes of games, a storage medium and an electronic device. The method comprises the following steps: acquiring request information of a first account, wherein the request information is used for requesting for inquiring fault causes of a first fault in a target game, and the first account is the account allowing to be usedin the target game; by responding to the request information, acquiring a matching result obtained after matching operation is carried out on the first account, wherein the matching operation is usedfor indicating and searching that whether a second account with the feature matched with that of the first account exists or not, and the second account is the account allowing to be used in the target game and used for requesting for inquiring the fault causes of the faults in the target game; and based on the matching result, searching for the fault causes of the first fault from a database, returning the fault causes of a second cause determined according to the causes of the first fault to the first account as the fault causes of the first fault, wherein the information of the corresponding fault causes is stored in the database. With the technical scheme, the technical problem that in the related technologies, the accuracy in locating the fault causes of the games is low is solved.

The invention discloses an association input method and an electronic device. The association input method comprises the steps of obtaining a first character string displayed on a screen by a user, and a second character string input after the first character string is displayed on the screen; and according to the first character string and the second character string, generating association candidate character strings, and displaying the association candidate character strings for the user to select and perform display on the screen. According to the technical scheme, association conversion is performed by utilizing a context relationship during input through combining contexts input by the user to generate more accurate association candidate character strings, so that the technical problem of relatively low accuracy during association and prediction input in the prior art is solved, the accuracy of association and prediction input is improved, and the input experience of the user isimproved.