48 results about "Character group" patented technology

The invention discloses a relation enhancement type named entity recognition method and device, a medium and a program product, and relates to the field of data processing. The method comprises the following steps: performing feature extraction on text data through a BERT model to obtain a plurality of tokens and semantic feature vector sequences; performing different-scale convolution operations by using a convolutional neural network to obtain local character combination features, and fusing the local character combination features with the semantic feature vector sequence to form a feature enhanced semantic feature vector sequence; the Transform model determines a long-distance dependency relationship of semantic feature vectors in the sequence by using an attention mechanism; generating a long entity structure model based on the long-distance dependency relationship, and then determining an enhanced semantic feature vector sequence; and performing tag sequence optimization processing on the named entity through a conditional random field layer, determining sequence labeling loss, boundary detection loss and relation prediction loss, determining total loss, and adjusting a tag sequence so as to determine the named entity. According to the method, the accuracy and integrity of named entity recognition can be effectively improved.

The invention discloses a training subject-oriented flexible configuration method and system, and relates to the technical field of training subject image transmission, and the method comprises the steps: periodically carrying out the image collection of a plurality of training actions of target trainees of a target training subject, and obtaining a periodic training image set; carrying out conversion processing on communication processing data converted from the training image set, constructing a first protection dictionary and a second protection dictionary in the conversion processing process, carrying out 1 judgment on each character in any character combination, obtaining a bit sequence based on a judgment result, selecting a related dictionary type according to the bit sequence, and carrying out data processing on the selected dictionary type; according to the method and the device, the key is encrypted to find the position of the element corresponding to the key to obtain the corresponding protection conversion sequence, so that the risk that the key is lost in the frequent transmission process due to the encryption mode of the key is avoided, the training atlas can be restored and obtained only through the specific dictionary and judgment logic, and a guarantee is provided for the transmission safety of the training atlas.

In a method and device for speech synthesis, the method of outputting text input as sound from an electronic device, includes receiving a text input including characters from at least two languages and at least one symbol, generating a plurality of ordered groups by sequencing, wherein the plurality of ordered groups includes character groups including characters from a common language, and a symbol group including symbols, by segmenting the text input sequentially by language and by symbol, generating sound segments corresponding to each of the character groups by use of a plurality of text-to-speech (TTS) engines supporting different languages, generating an output sound by merging the sound segments according to the sequencing of the character groups within the plurality of ordered groups, and displaying the symbol of the symbol group on a display while outputting the output sound by use of a speaker.

An object of the present invention is to enhance a player's motivation regarding a game result in relation to a character object. An information processing program is implemented on a computer: a first game control function for running a first game when enabling a character group to participate in the first game; a subset creation function for creating, based on the game result of the first game, a subset composed of a part of the characters included in the character group from the entire set composed of all the characters included in the character group; and a second game control function for running a second game when controlling the characters included in the subset created by the subset creation function so that at least a part of the character objects are displayed on a display medium in the form of a predetermined type of character object.

The invention discloses an embedded host data encryption method and encryption system in the technical field of data encryption. The method comprises the following steps: setting a data transmission starting password and a 16-bit character index table of a microprocessor; correspondingly converting the data transmission starting password into a binary ASCII value according to an ASCII code table; performing grouping calculation on the binary ASCII values to obtain a decimal character array, sequentially converting each value in the decimal character array according to a 16-bit character index table to generate a symbol ciphertext character group of a data transmission starting password with N values, and taking the value of the symbol ciphertext character group as a first-level symbol ciphertext character; and sequentially carrying out logical operation on the ASCII code values of the adjacent values in the symbol ciphertext character group. According to the method, the 16-bit character index table is set, so that the data transmission starting password can be encrypted into two kinds of ciphertext character strings, namely the first-level symbol ciphertext character and the second-level symbol ciphertext character, and the leakage risk of the data transmission starting password of the embedded host can be effectively reduced.

A control portion recognizes, as an independent character region, a region of which the absolute value of the difference between the width in a first direction, which is the writing direction, and the width in a second direction, which is orthogonal to the first direction, is smaller than a first threshold value, and checks, based on the width in the first direction of a reference region that is a character region adjacent, in the second direction, to a plurality of independent character regions aligned in the first direction but that is not an independent character region, whether a character string composed of characters in the plurality of independent character regions is one word. On judging it to be one word, the control portion deals with the character string resulting from uniting the characters in the plurality of independent character regions as one word.

The present application discloses a method, apparatus, device and medium for intent recognition, which includes: obtaining a training sample set, the training sample set including multiple training samples, each training sample being a character combination of an abstract intent, and the training sample including at least ordered arranged characters of interest; when the total number of characters in the training sample set after deduplication is greater than or equal to a first threshold, compressing each training sample according to the number of character appearances to obtain a compressed sample set; inputting the compressed sample set into a pre-constructed masked language model for training, and outputting an intent recognition result, the masked language model being used to fill the input compressed sample into a preset position in a language prompt template to be trained and then train. The embodiment provided in the present application effectively improves the convergence speed of the masked language model by uniformly pre-processing the training samples.

The invention discloses an image optimization method and system combined with an attention mechanism, relates to the technical field of image optimization, and solves the problems that global unified parameter adjustment (such as overall sharpening) is often adopted, differentiation processing cannot be carried out on a target area (such as a character main body) and a background area, and the image optimization efficiency is low. According to the method, based on a preset weight coefficient (for example, when Query is a person, A1 = 0.685 focuses on the number of inflection points), process feature calculation (Value value) is carried out on a combination of similar features, and an optimal combination is screened by maximizing process features; the mechanism can inhibit interference of non-target features (such as dogs in the background), ensure accurate locking of to-be-optimized features (such as character groups), and realize efficient calculation of Value values in the attention mechanism.

The application provides a special-shaped character detection and recognition method, system, device and storage medium, and relates to the field of special-shaped character detection and recognition.The method comprises the following steps: (1) constructing a training set based on an existing product image, and importing the training set into a rotating target detection model for training; (2) inputting a product image to be detected into the trained rotating target detection model to output string information M of different regions in the product image to be detected; and (3) based on the string information M of different regions, adopting an optimal combination algorithm to combine characters in front and back, and forming a plurality of independent strings.The application uses a rotating target detection model for character recognition, and adopts an optimal combination algorithm for character combination on the basis of character recognition, and then detects special-shaped strings on medical devices and other products.The application solves the problem of low character detection accuracy caused by irregular string shapes in a complex scene.

The invention discloses a character rendering method for character combination frequency priority and texture splicing processing. The method comprises the following steps: collecting character combinations in a rendering scene and counting the occurrence frequency of the character combinations; the method comprises the steps of generating bitmap data of a single high-frequency character of a high-frequency character combination according to the high-frequency character combination, pre-splicing the bitmap data of the high-frequency character according to character position information to generate a large texture data graph, preferentially creating a texture object of the high-frequency character combination and storing the texture object in a video memory, calculating and storing texture coordinates of the high-frequency character combination during texture loading, and establishing a mapping relation between the high-frequency character combination and the texture coordinates thereof; carrying out generalization processing on the low-frequency character combination; and based on the loaded high-frequency splicing texture and the dynamically loaded low-frequency texture, finishing drawing of the character combination in combination with texture coordinates. The number of times of texture binding, switching and drawing in character rendering is reduced, and the character rendering efficiency and system performance of a display system are improved.

The invention discloses a metering detection data digital interaction method and system, and relates to the technical field of metering interaction.The method comprises the steps that after metering collection data of target power equipment is obtained, a reference conversion sequence used for conversion is constructed, and conversion processing is conducted on the real-time metering collection data according to the reference conversion sequence; according to the construction of the reference conversion sequence, character composition types of four-bit binary numbers of numeric characters from 0 to 15 serve as the basis, the types of characters removed from the reference conversion sequence are determined according to the frequency quantity of different two-bit binary numbers in each mapping sequence, so that a new reference conversion sequence is obtained, and each new reference conversion sequence is obtained. When the measurement detection data is transmitted, the measurement detection data and the mapping sequences are subjected to AND operation, so that the corresponding transformation sequences are obtained, the dynamism and complexity of each group of mapping sequences are guaranteed, the reference transformation sequences which are subjected to AND operation with each group of mapping sequences are in dynamic change, and the safety of the measurement detection data in the transmission process is further guaranteed.

The invention discloses an interaction method and device, equipment and a storage medium, and relates to the technical field of Internet. The method comprises the following steps: displaying a virtual environment, wherein the virtual environment comprises a first virtual character; under the condition that the first virtual character moves to a first area of the virtual environment and the character group holding the cruise performance activity is located in the first area, the cruise performance activity held by the character group in the first area is displayed, and the character group walks to different areas of the virtual environment to hold the cruise performance activity; and in the holding process of the cruise performance activity, in response to the first user account, controlling the first virtual character and the character group to complete interaction, and displaying a first interaction reward obtained by the first user account. In the role group cruising process, interaction with the role group is supported to obtain the interaction reward, so that the interaction mode is not fixed and single any more, the variability and diversity of interaction are improved, the probability of user interaction is improved, and the interactivity of the application program is improved.

This invention discloses a method for recognizing pointer-type instrument dial information based on object detection, comprising: 1) extracting scale digits and pointer information from the instrument dial image using an object detection network; 2) performing non-maximum suppression processing on characters with different labels; 3) dividing adjacent individual characters into character groups to be fused; 4) generating scale digit information based on the individual character information within the character groups to be fused; 5) estimating the pointer rotation center point using the scale digit information; 6) applying an improved adaptive Hough transform line detection method to the pointer region to fit the line where the pointer is located; and 7) calculating the coordinates of the pointer tip based on the pointer rotation center point and the line where the pointer is located. This invention utilizes an object detection network, which can establish complete dial information based on each individual character, and adaptively analyze the pointer region to obtain pointer information. It has good versatility, detection speed, and accuracy, and can effectively improve the performance of instrument reading algorithms.

To enable a user to save labor required for selecting an input form suitable for character strings desired for printing from a plurality of options.SOLUTION: A tape printer includes an acquisition part for acquiring image data, a storage part for storing a storage character string as one or more characters and an input form 80 having a character input column 83 for each printing item 82 in association with each other, a display 22, and a control part, wherein the control part image-analyzes the acquired image data and extracts a plurality of characters, and displays the input form 80 associated with the storage character string on the display 22, when the same character string as the storage character string stored in the storage part is included in an extraction character group including the plurality of extracted characters.SELECTED DRAWING: Figure 8

The invention relates to a handwritten mathematical expression recognition method based on a skeleton shaping and character counting module, and belongs to the technical field of deep learning and image content recognition. The method comprises the following steps: firstly, constructing a framework of an encoder-decoder model, adopting DenseNet as an encoder for extracting features from an image, adopting Transform as a decoder, and capturing a dependency relationship in an input sequence through a self-attention mechanism; secondly, a skeleton shaping and character counting module is provided, the module predicts character number information through a character frequency distribution prediction module, and expression structure information is extracted through a skeleton shaping device; then, the structure information and the decoded features are integrated in the structure calibration module, so that the model can determine the spatial relationship of the sub-structures in the sequence according to the positions of the structure characters. And finally, in the number calibration module, the model can use the known character number to infer the most possible character combination, and outputs a final prediction result.

The present application discloses an intelligent input method, device, and storage medium. The method includes: identifying an acquired character string to obtain at least one target character group, and obtaining a group serial number corresponding to each target character group; calculating multiple semantic probability values ​​of the character string based on at least one semantic corresponding to the group serial number and a reference text, and taking the semantic represented by the highest semantic probability value as the target semantic of the character string, the reference text including input records and / or chat records; inputting the target semantic into a sentence deduction model, which generates at least one initial sentence and an applicable probability value for each initial sentence based on the target semantic; sorting the initial sentences based on the applicable probability value, and displaying the top K initial sentences as target sentences, where K is a positive integer. The intelligent input method described in the technical solution of the present application generates input content that fits the user's input habits while facilitating the user's input operation.

The method, device, storage medium, and computer equipment for determining the text recognition result provided by this application can, when performing OCR recognition on the text image to be recognized, first determine the character corresponding to the extremely high probability value in the probability distribution of each time step, and then, for each time step, determine whether the character corresponding to the extremely high probability value in the probability distribution of this time step is a character in the pre-configured error-prone character grouping list. If not, directly use the character corresponding to the extremely high probability value in the probability distribution of this time step as the output character of this time step; if so, update the extremely high probability value based on the probability value of the error-prone character paired with the character corresponding to the extremely high probability value in the error-prone character grouping list in the probability distribution of this time step, and use the character after updating the extremely high probability value as the output character of this time step. In this way, the situation where the overall score is too low due to the ambiguity of easily confused characters can be avoided, thereby effectively improving the accuracy of the recognition result.

The invention relates to a control method, device and equipment of an intelligent door lock and a storage medium. The intelligent door lock comprises a plurality of key character groups, each key character group comprises a plurality of key characters, the positions of the key characters are represented by character arrangement, different key character groups correspond to different display areas, and the method comprises the following steps: in response to an unlocking request operation of a user, determining a target key character group from the key character groups, the target key character group is selected, the target character arrangement of the target key character group is determined, the target character arrangement is different from at least one historical character arrangement, and the historical character arrangement is the character arrangement historically presented by the historically displayed key character group; and in a display area corresponding to the target key character group, displaying the target key character group according to the target character arrangement. According to the embodiment of the invention, the risk that a correct password is cracked by an attacker can be reduced, and the user experience can be improved due to the fact that multiple virtual passwords do not need to be input.

This invention provides a method and apparatus for Manchu speech synthesis based on character component modeling, belonging to the field of speech synthesis technology. The method includes: converting Manchu text into Latin characters according to transcription rules, generating a character embedding sequence, and generating feature templates containing dimensions such as fundamental frequency and pitch band based on the acoustic characteristics of Manchu pronunciation; concatenating the two templates into a high-dimensional input tensor, which is then fed into a pre-trained duration prediction network; constructing a mask, extracting prosodic features from the Manchu Latin-transcribed character embedding sequence to generate a prosodic vector, and adjusting the attention weights between characters and speech frames in the pre-trained feature network to obtain the prosodic space, Mel-spectral sequence, and optimal alignment path; finally, inputting the vector into a lightweight multi-band inverse short-time Fourier transform decoder for Manchu to obtain a complete synthesized speech waveform. This invention can significantly and effectively solve the core problem of text-speech frame matching deviation in low-resource Manchu scenarios, and its inference speed is significantly improved compared to general models.

[Problem] To ensure that communication elements are amusing. [Solution] A game device comprising: a storage means for storing information of a character group; an evaluation means for receiving evaluation, by a user, of each character in the character group the information of which has been stored in the storage means; an accumulation means for accumulating the evaluations concerning the character group received by the evaluation means; and a first transmission means for collectively transmitting the evaluations accumulated in the accumulation means to a server.

The present invention relates to the technical field of data encryption processing, and in particular to a method for intelligent management of trajectory calculation data, the method comprising: obtaining data before and after calculation by trajectory software as original data to be managed; selecting data from the original data as initial data; obtaining a character combination sequence and a temporary index table of the initial data, and counting the frequency of occurrence of each character combination in the character combination sequence in the original data as the influence degree of the corresponding character combination; obtaining new characters in the original data, and obtaining an updated character combination based on the new characters and the influence degree of each character combination; disassembling the updated character combination to obtain at least two decomposed character combinations, and obtaining a secondary index sequence of the updated character combination in combination with the original data to obtain an optimized index table; encoding and encrypting the original data according to the optimized index table to obtain ciphertext data for management, thereby improving the data encryption effect and information hiding, and ensuring data security.

An information processing method, device, equipment, storage medium and program product are disclosed, the method comprising: obtaining a to-be-detected target sentence of a user, the to-be-detected target sentence being composed of multiple characters; determining corrected characters according to the multiple characters through an error correction model; the error correction model being obtained by training a pre-trained language model, a semantic aggregation layer network and a classifier, the pre-trained language model being used to generate character vectors, the semantic aggregation layer network being used to generate error correction vectors according to the character vectors, and the classifier being used to generate the corrected characters according to the error correction vectors; wherein the corrected characters compose a corrected sentence, which is used to support voice interaction with the user. The present application can solve the problem of complex processing and low accuracy of error detection in the prior art voice error correction method.

The invention provides a moire removal method and device for an image, a medium and electronic equipment, and the method comprises the steps: carrying out the preprocessing and recognition of a to-be-removed image through employing an OCR recognition technology, and precisely positioning the position and type of each target character. And according to the distance between the adjacent target characters, reasonably grouping the target characters to form a target character group. The moire pattern influence areas are accurately positioned in a grouping mode, a proper moire pattern removing algorithm is used in a targeted mode, unnecessary transition blurring of the whole image is avoided, and the detail definition of the areas which are not influenced by moire patterns is reserved. Therefore, compared with an existing scheme, the moire removal effect is more accurate, and the problem that the moire removal effect of the image in the existing scheme is poor is solved.

The invention provides a full-language speech synthesis method and device based on character component modeling, and relates to the technical field of speech synthesis. The method comprises the following steps: converting Manchu into Latin characters according to a transcription rule, generating a character embedding sequence, generating a feature template containing dimensions such as a fundamental frequency and a pitch band based on full-language pronunciation acoustic characteristics, splicing the feature template and the feature template into a high-dimensional input tensor, and sending the high-dimensional input tensor into a pre-training time length prediction network; constructing a mask, extracting rhythm-related features from a character embedding sequence of Manchu Latin transcription to generate a rhythm vector, adjusting attention weights of characters and voice frames in a pre-training feature network to obtain a rhythm space, a Mel spectrum sequence and an optimal alignment path, and finally inputting the rhythm space, the Mel spectrum sequence and the optimal alignment path into a full-language lightweight multi-band inverse short-time Fourier transform decoder. And a complete synthetic voice waveform is obtained. According to the method, the core problem of text and voice frame matching deviation in a full-language low-resource scene can be obviously and effectively solved, and the reasoning speed is obviously improved compared with a general model.

The present invention particularly relates to a character grouping method based on bigram frequencies and a method for constructing a secure character library. The character grouping method includes the following steps: traversing a corpus to statistically obtain a bigram frequency matrix of the occurrences of any two characters among N characters to be grouped; traversing the characters one by one in descending order of character frequency, and calculating the weight of the character c to be assigned to the k-th group according to a formula; adding the character c to be assigned to the group with the largest weight, and so on until all characters are grouped. The bigram frequency matrix reflects the frequency of two characters appearing together. Through the weight calculation formula, the weight increases when two characters that often appear together are in different groups. In this way, we can make the characters that appear together be in different groups as much as possible by selecting the group with the largest weight, thereby realizing reasonable grouping of characters. This grouping method does not limit the number of characters in each group, making it more reasonable.

The application discloses an audio-based device network configuration method and system, a storage medium and an electronic device. The method comprises the following steps: obtaining first network configuration information to be transmitted to a device to be configured, the device to be configured being a device to be configured, and the first network configuration information being a to-be-transmitted character sequence, each to-be-transmitted character being a preset character in a preset character group; selecting a target frequency group from a plurality of preset frequency groups corresponding to the preset character group according to the noise frequency of the current environmental noise, the frequency segments corresponding to different preset frequency groups being non-overlapping, and different preset characters corresponding to different preset frequencies in each preset frequency group; generating an audio signal corresponding to each to-be-transmitted character according to a target frequency corresponding to each to-be-transmitted character in the target frequency group in the to-be-transmitted character sequence; and playing the audio signal corresponding to each to-be-transmitted character in sequence through an audio playing component to transmit the first network configuration information to the device to be configured.

The invention relates to the technical field of computer vision and optical character recognition, and discloses a clustering-based character component matching method and system. The character component matching method based on clustering comprises the following steps: performing optical character recognition processing on a to-be-processed image to obtain a plurality of characters; the method comprises the following steps: firstly, performing spatial clustering processing on all characters to obtain at least one initial space group, and then performing angle clustering processing on the characters in each initial space group to obtain at least one character subgroup; performing projection reconstruction on the character subgroup, and determining the arrangement sequence of all characters; sorting all the character subgroups according to the sorting sequence of all the characters to obtain an ordered character string sequence; according to the method, the discrete OCR recognition result is reconstructed into an ordered character string sequence by combining double clustering with a dynamic projection reconstruction technology, so that the character sequence reconstruction accuracy and the automation degree are effectively improved.

The invention provides a license plate number recognition method, device and equipment based on an entity extraction technology, and the method comprises the steps: carrying out the structural processing of a first text, so as to obtain a second text, and enabling the first text to be obtained through the text transliteration of a call record between a seat and a customer; character recognition and replacement are conducted on to-be-processed characters in the second text on the basis of a preset mapping dictionary to obtain a third text, the preset mapping dictionary is used for indicating the mapping relation between the to-be-processed characters and the target form characters, and the to-be-processed characters are homophonic characters or Chinese form numeric characters; the target form characters are determined based on license plate number naming rules; and performing matching processing on the third text based on a first regular expression to identify and determine a target character group, the first regular expression being constructed based on a license plate number naming rule. Therefore, the homophonic characters or Chinese-form numeric characters in the text can be effectively processed, so that the accuracy of license plate number recognition is effectively improved.

To provide an image processor, an image processing system, an output device, an image processing method, and an image processing program for improving the character recognition accuracy of a character string including a small-sized handwritten character.SOLUTION: The image processing apparatus 1 includes an acquisition processor 111 that acquires character image data, a character string extraction processor 112 that extracts a character string rectangle corresponding to a character string composed of a plurality of characters from the character image data, a one character extraction processor 113 that extracts a plurality of one character rectangles respectively corresponding to the plurality of characters from the character image data, a correction processor 114 that corrects a specific one character rectangle when the plurality of one character rectangles include the specific one character rectangle in which at least one of a position and a size satisfies a predetermined condition, and a recognition processor 115 that executes character recognition processing of the character string using the specific one character rectangle corrected by the correction processor 114.SELECTED DRAWING: Figure 1

The invention relates to the technical field of business information consultation, and discloses a data storage method of a business information consultation system, comprising: S1, data acquisition and analysis; s2, character frequency statistics; s3, character combination generation; s4, influence factor calculation; s5, generating a character list; s6, coding compression; and S7, carrying out distributed storage. According to the data storage method and system of the business information consultation system, after a data acquisition module analyzes dialogue data to generate a character sequence, a frequency analysis module counts a first probability, and a character combination processing module extracts a high-frequency three-character combination by using a sliding window and combines the high-frequency three-character combination into a multi-level set; the influence factor calculation module adjusts a second probability in combination with co-occurrence and global frequency, the list generation module generates a character list according to a front probability sequence, and the coding compression module dynamically adjusts the sequence by moving to a front-end algorithm, so that the coding quantity is reduced, the compression efficiency is improved, and the occupied storage space is reduced.