Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

27results about How to "Guaranteed readability" patented technology

Industrial large model-oriented fine tuning data generation method, computer program and terminal

The invention discloses a fine tuning data generation method for an industry large model, a computer program and a terminal, and belongs to the field of artificial intelligence, and the method comprises the following steps: extracting text data in a professional data source picture; preprocessing the text data; identifying and correcting the grammar of the text data; calculating the confusion degree of the text data to eliminate sentences with low scores; and segmenting the text data, inputting the segmented text data into the large language model, and generating fine tuning data in combination with prompts. A text recognition model is adopted to recognize text data in a professional data source picture, and the problem of large model illusion is avoided; the grammar and spelling errors of the text data are checked and corrected by using the large language model enhanced through reinforcement learning, so that the data generation quality can be further improved; the text data are segmented and then input into the large language model, the method is more suitable for large-scale text data processing, semantic boundaries in the text can be effectively recognized while high efficiency is kept, and high-quality instruction fine adjustment data are generated.
Owner:UNIV OF ELECTRONICS SCI & TECH OF CHINA

Lyric and rhythm recognition printing method based on audio fingerprints and related equipment

The invention discloses a lyric and rhythm printing method based on an audio fingerprint and related equipment, and the method comprises the steps: carrying out the preprocessing operation of a to-be-processed audio, and obtaining the preprocessed audio data; dividing the preprocessed audio data into a plurality of segments; respectively extracting key features of each fragment; converting each key feature into a corresponding audio fingerprint through a Hash algorithm; searching audio fingerprints in a preset song library and judging whether matching succeeds or not; if the matching is successful, first primary printing information is generated, the first primary printing information comprises lyric information and rhythm information arranged according to a preset format, and the lyric information and the rhythm information are lyric information and rhythm information corresponding to the matched audio fingerprint; the first primary printing information is rendered according to a preset style, and a final printing picture is obtained; and sending the final printing picture to the printing equipment to enable the printing equipment to print the final printing picture. The lyrics and rhythms can be quickly and accurately recognized and printed, and the accuracy and efficiency are improved.
Owner:SHENZHEN EFERCRO ELECTRONIC TECHNOLOGY CO LTD

Text expansion method and device, electronic equipment and storage medium

The embodiment of the invention provides a text expansion method and device, electronic equipment and a storage medium, and belongs to the technical field of text processing. The method comprises the following steps: acquiring an original text sample and a reference text sample; word embedding is carried out on the original text sample based on the initial text expansion model to obtain word embedding features; performing position random embedding on the original text sample based on the initial text expansion model to obtain position embedding features; performing style attention coding on the word embedding features and the position embedding features based on an initial text expansion model to obtain style text coding features; decoding the style text coding features based on the initial text expansion model to obtain an original expansion text; training the initial text expanding-writing model based on the original expanding-writing text and the reference text sample to obtain a target text expanding-writing model; and performing text expansion on the target original text based on the target text expansion model to obtain a target expansion text. According to the embodiment of the invention, the readability of the text generated by expanding writing can be improved.
Owner:PING AN TECH (SHENZHEN) CO LTD

High-temperature-resistant primer, application thereof and marking composition for honeycomb ceramics

The invention provides a high-temperature-resistant primer and application thereof, and a marking composition for honeycomb ceramics, and relates to the technical field of honeycomb ceramic surface coatings, the high-temperature-resistant primer comprises the following components by mass: 35%-50% of colored particles, 3%-5% of an emulsifier, 5%-15% of epoxy resin, 5%-10% of a thickener, and 15%-20% of an organic solvent; wherein the colored particles include metal alkoxide particles; the primer is cured on the surface of the honeycomb ceramic to form a coating, and the identification information is marked on the coating, so that a clear and stable identification effect of the identification information can still be kept under a high-temperature condition, and the technical problem that the identification information is difficult to clearly display on a honeycomb ceramic primer layer in a high-temperature environment in the prior art is solved.
Owner:SHANDONG SINOCERA FUNCTIONAL MATERIAL CO LTD

Automatic plotting method and system based on traffic field long and large linear engineering BIM model

The invention provides an automatic drawing method and system based on a BIM (Building Information Modeling) of a long and large linear project in the traffic field, and the method comprises the steps: obtaining a two-dimensional view based on the BIM, and carrying out the marking and releasing of the model in the two-dimensional view, and constructing a complete view; designing a statistical table template based on actual demands, and extracting BIM model information to fill statistical table template content to generate a statistical table; a drawing template is designed based on actual demands, wherein the drawing template comprises a mapsheet, a drawing frame, a countersign column, a project information column and / or a drawing area; selecting a required drawing template, performing layout on the drawing area to construct a view template, and obtaining a required two-dimensional view and / or a statistical table based on the view template to generate a complete drawing; drawing output parameter setting is conducted on the complete drawing, output content is selected, printing configuration parameters are set, and the drawing is printed and output. According to the method, the drawing efficiency can be greatly improved, human resources are saved, manual errors are reduced, the readability of the drawing is enhanced, and a reader can quickly position effective information conveniently.
Owner:CHINA TRANSPORT INFORMATION TECH GRP CO LTD

Method and system for dynamically extracting information from electronic checklist forms

ActiveCN117391059BGuaranteed readabilityReduce workloadText processing
The present invention discloses a method for dynamically extracting information from an electronic checklist (ECL) form, comprising: detecting whether a form exists in an ECL program; if a form exists in the ECL program, determining whether information extraction from the form is required; in response to determining that information extraction from the form is required, determining target parameters based on the content of the form; determining an extraction range based on aircraft status information and the target parameters; and extracting information from the form based on the extraction range to display the extracted information in the ECL program.
Owner:COMMERCIAL AIRCRAFT CORP OF CHINA LTD +1

LED structure and LED screen thereof

The invention relates to the technical field of LEDs, in particular to an LED structure and an LED screen thereof.The LED structure comprises a substrate, a plurality of bulbs are fixedly installed on the substrate in an array mode, the substrate and the bulbs are jointly provided with compensation devices, each bulb is provided with one compensation device, the compensation devices guide light emitted by the bulbs in a reflection mode, and the light emitted by the bulbs is reflected by the compensation devices. A light gathering device is arranged on the substrate, the light gathering device is used for gathering light of a plurality of bulbs to the same compensation device, the compensation device comprises a mounting plate fixedly mounted on the bulbs, an optical fiber is fixedly mounted on the mounting plate, and one end of the optical fiber extends out of one side, far away from the substrate, of the mounting plate; and a reflecting part I is arranged on one side, close to the substrate, of the mounting plate. The invention provides an LED structure and an LED screen thereof so as to improve readability and accuracy of information displayed by the LED screen after local LED bulbs of the LED screen are damaged.
Owner:潘璐璐

JAVA-based IVR (Interactive Voice Response) framework implementation method and system

The invention provides an IVR (Interactive Voice Response) framework implementation method and system based on JAVA, a VXML 2.0 protocol is taken as a basis, a form of a custom XML (Extensible Markup Language) label is packaged through JAVA according to a service scene, and the custom XML label is packaged with a VXML 2.0 standard function and a service logic to realize various functions in an IVR service; in the telephone traffic request initialization scene, the IVR service platform triggers a service request through an HTTP interface request mode; in a business process assembly scene, arranging and assembling the packaged self-defined labels according to business scene requirements; and in the business analysis scene, the assembled XML document is output, and the IVR service platform carries out analysis operation to complete related telephone traffic actions, so that the rapid completion of the IVR telephone traffic process is realized. According to the invention, the service calling function of the IVR process can be realized in a relatively universal manner. And application developers only need to have a JAVA basis, so that the difficulty of application development is reduced, and the research and development quality is also ensured.
Owner:CHINA YOUKE COMM TECH

Typesetting method and typesetting equipment capable of shooting and typesetting at same time, medium and program product

The invention discloses a typesetting method and typesetting equipment capable of printing while shooting, a medium and a program product, and relates to the technical field of image processing and intelligent printing. On the basis of converting an original image into structured data and performing global mapping, the typesetting device divides consumables into a main visual area, an auxiliary visual area and an avoidance area, constructs a semantic adhesion cluster in combination with spatial adjacency between objects, and can detect whether an image-text module with close logic association can be split by a fold line or shielded by a hole. If conflicts occur, the limitation of global scaling is broken, and local scaling and offset coordinates are independently calculated for the adhesion clusters for adjustment. According to the derivation process of multi-dimensional feature interaction, the problems of visual blind areas and semantic segmentation of traditional two-dimensional typesetting in a complex three-dimensional pasting scene are effectively avoided, and the integrity and transmission accuracy of graphic and text information of special labels (such as flag-shaped labels and winding labels) in actual physical application are improved.
Owner:BEIJING SHUOFANG INFORMATION TECH CO LTD

Industrial innovation resource data analysis method and system

The present application provides an industrial innovation resource data analysis method and system, which performs dimensionality reduction processing on the network resource interaction data to be analyzed, thereby obtaining the reduced-dimensional network resource interaction data, and then determines it as interactive event sample data in combination with the reduced-dimensional network resource interaction data, and then analyzes the sample set corresponding to the above interactive event sample data in the network resource interaction data to be analyzed to determine whether each sample set is the first interactive event, that is, to determine the interactive event analysis results in the network resource interaction data to be analyzed, and the interactive event analysis results are the location to which the interactive event belongs. Since the interactive event sample data is selected through dimensionality reduction processing, it is only necessary to perform significance analysis on the sample set that matches the interactive event sample data, which reduces the difficulty of analysis, thereby improving the accuracy and credibility of the analysis, and therefore, can ensure the integrity and readability of the interactive event analysis results.
Owner:GUANGZHOU DOCTOR INFORMATION TECH RES INST CO LTD

Structure-improved multi-hole thread packaging machine

The utility model discloses a multi-hole thread binding machine with an improved structure in the field of thread binding machines, which comprises a workbench, a punching mechanism and a paper pressing mechanism, wherein the workbench is used for bearing external parts to be bound, and the punching mechanism and the paper pressing mechanism are arranged on the workbench. The punching mechanism comprises a drilling tool, a first driving assembly for driving the drilling tool to punch a to-be-bound piece, and a collecting assembly for collecting paper scraps; the paper pressing mechanism comprises a paper pressing plate and a second driving assembly for driving the paper pressing plate to limit the to-be-bound piece on the workbench, an inductive switch matched with the second driving assembly for use is arranged on the paper pressing plate, and an inductive piece of the inductive switch protrudes out of the paper pressing plate. The induction piece touches the to-be-bound piece or the workbench so that the induction switch can be matched with the second driving assembly to brake the paper pressing plate. Through the additional arrangement of the inductive switch, the phenomena of folding marks, tearing and the like of the to-be-bound piece caused by continuous pressurization of the paper pressing mechanism are avoided, and the binding quality of the thread binding machine is improved.
Owner:GUANGDONG PIAOYOU INTELLIGENT TECH CO LTD

A method and system for automated evaluation of robustness of a text variation-based classification model

PendingCN122594491AAssess sensitivityEnsure consistency
The application belongs to the field of natural language processing and artificial intelligence security evaluation, and specifically discloses a kind of classification model robustness automatic evaluation method and system based on text variation, comprising: obtaining text data set with original classification label and carrying out uniform mapping to heterogeneous label;Compound text variation is carried out for each sample text, and the core processing mechanism is to convert the implicit alias of the evaluation object noun, and the consistent replacement constraint is maintained in the same sample text;Then input the original text and the variation text into the same text classification model, compare the consistency between the corresponding prediction label and the original label of the two, and calculate the label retention ability and robustness evaluation index of the model under the condition of text variation.The application can evaluate the degree of dependence of the model on explicit object address, strong emotional words and fixed expression templates without changing the basic understandability of the text, and is suitable for robustness analysis of various text classification tasks based on artificial intelligence model.
Owner:UNIV OF ELECTRONICS SCI & TECH OF CHINA

Abdomen image display equipment for radiology department

The invention discloses abdomen image display equipment for the radiology department, and relates to the technical field of medical equipment. The abdomen image display equipment for the radiology department comprises a displayer arranged on a workbench and further comprises a protection mechanism arranged on a shell and used for preventing atomization of a display screen, the protection mechanism comprises a protection pipe fixedly connected to the side, away from the display screen, of the shell, and the side, away from the workbench, of the protection pipe is fixedly connected with a plurality of L-shaped pipes; and the other end of each L-shaped pipe is arranged opposite to the display screen, and an air heater is arranged on the side, close to the protective pipe, of the shell. According to the abdomen image display equipment for the radiology department, through the arrangement of the protection mechanism, the surface temperature of the display screen is properly increased, evaporation of moisture on the surface of the display screen is accelerated, and the humidity is reduced, so that the possibility of atomization is reduced, and the definition and readability of image display of the display screen are guaranteed; and thus, the accuracy of diagnosis of the image by the doctor is improved.
Owner:河北港口集团有限公司秦皇岛中西医结合医院

A blockchain traceability data management platform for digital agriculture

The application belongs to the technical field of digital agriculture, and particularly relates to a blockchain traceability data management platform for digital agriculture, wherein the system comprises an agricultural product supply chain informatization module, a trust value quantification module, a two-dimensional code generation module, a local two-dimensional anti-fraud verification module and a traceability query module. The scheme constructs a distributed blockchain ledger network suitable for the agricultural product supply chain, allocates unique on-chain identity and permission to each link participant, generates a basic two-dimensional code bearing core data through a two-dimensional code generation library, triggers a smart contract to calculate a trust value in real time and uploads the trust value to the chain during traceability query, calculates a local hash value by extracting an on-chain identity embedded in a two-dimensional code to be detected, and compares the local hash value with a preset authorization list to complete legality verification, realizes real-time two-dimensional anti-fraud verification of the two-dimensional code without network dependence, can accurately trace and immediately intercept a counterfeit behavior, and provides a full-process anti-counterfeiting and traceable trust guarantee for agricultural product traceability.
Owner:HUNAN INST OF INFORMATION TECH

U-shaped automatic wharf intelligent guiding device

The invention discloses a U-shaped automatic wharf intelligent guiding device, and belongs to the technical field of wharf equipment, the U-shaped automatic wharf intelligent guiding device comprises a bottom plate, an outer sleeve is fixed on the side surface of the top end of the bottom plate, a locking and inserting assembly and a guiding assembly are arranged between a sliding support rod and the outer sleeve, a connecting rod is fixed at the bottom of a line navigator, and an inclination adjusting assembly is arranged between the connecting rod and the sliding support rod. The top end of the line navigator is provided with a cleaning assembly. According to the line navigator, the angle of the line navigator can be adjusted through the inclination adjusting assembly, a display screen on the line navigator can be adjusted to a proper angle, the sliding supporting rod in the outer sleeve is locked through the locking and inserting assembly, and therefore the height position of the line navigator can be adjusted; therefore, the display screen can be suitable for wharf control operators of different heights to observe and read information, the display screen is blown and cleaned through the cleaning assembly, the definition of the display screen is ensured, the cleaning convenience is effectively improved, and the service life of the display screen is prolonged.
Owner:GUANGXI QINZHOU FREE TRADE PORT ZONE SHENGGANG TERM CO LTD

Method of programming memory, memory device, and memory system

This disclosure provides a memory programming method, memory device, and memory system. The method includes: performing a first programming on each memory cell to be programmed, such that each memory cell is programmed into N different first programming states; wherein N is a positive integer greater than 1 and less than M, and M is the total number of target programming states for the memory cells to be programmed; performing a second programming on the i-th group of memory cells in the first programming states, such that the i-th group of memory cells in the first programming states is programmed into k target programming states; wherein at least two groups of memory cells in the first programming states obtain different values ​​of the number of target programming states k after the second programming; k is greater than or equal to 1 and less than N; i is greater than or equal to 1 and less than or equal to N.
Owner:YANGTZE MEMORY TECH CO LTD

An editor system and method for automatically drawing a structural formula based on a large language model

PendingCN122658504ANo need to learn how to operateAchieve precise reference
The application discloses a structural formula editor system and method based on automatic drawing of a large language model. The method comprises the following steps: acquiring, by a canvas context perception module, atom numbers, bond numbers and attribute information of a current molecular structure on an editor canvas; visualizing the atom numbers and the bond numbers of the molecular structure by an atom and bond visualization numbering module; inputting canvas state information and a user natural language instruction into a large language model; the large language model analyzes a user intention and generates a tool calling instruction corresponding to a skill; and a skill execution engine receives the tool calling instruction and executes a structure drawing operation on the editor canvas. The application realizes accurate drawing and editing of a chemical structure formula driven by a natural language dialogue.
Owner:QINGDAO KINGAGROOT CHEM COMPOUNDS CO LTD

Josephh ring concept-based equal-length symmetric encryption and decryption method and device and medium

The invention discloses an equal-length symmetric encryption and decryption method based on a Josephh ring concept, and the method comprises the steps: setting two random keys key1 and key2 which are respectively used for the conversion from a plaintext to a numerical value and the conversion from the numerical value to a ciphertext; setting a variable parameter step of the Josephh ring; encrypting the plaintext by using the Josephh ring parameter; and analyzing the ciphertext by using Josephh ring inverse conversion. The method has the beneficial effects that two complex and efficient modes of key conversion and Josephh ring conversion are combined, and the difficulty of ciphertext decoding is greatly improved. In this way, the security of the encrypted character string is ensured, and the information is effectively prevented from being illegally acquired and utilized. According to the method, encryption is carried out under the condition that the length of the original character string is not changed, it is ensured that the length of the encrypted ciphertext is completely the same as the length of the plaintext, the effectiveness and readability of information are ensured, and the transmission problem caused by the too long ciphertext is avoided.
Owner:SUZHOU REKTEC INFORMATION TECH CO LTD

Intelligent community property management system

ActiveCN121235858BGuaranteed readabilityRealize dynamic optimization configurationData processing applicationsData accessCommunity setting
The application relates to the field of intelligent community operation and maintenance scheduling, and discloses an intelligent community property management system, which comprises a structured data access and mapping unit, a dynamic confidence calculation unit, a space-time consensus fuse check unit and a space gravity field scheduling engine.The structured data access and mapping unit is used for converting the structured event label of a service request into a basic score; the dynamic confidence calculation unit is used for calculating a confidence coefficient according to the historical operation and maintenance interaction record of an initiating user; the space-time consensus fuse check unit is used for triggering logical fuse and setting the priority to the highest when detecting that the number of space-time associated requests exceeds a threshold value; and the space gravity field scheduling engine is used for locking the highest priority request as an anchor point and weighting and correcting the priority of secondary requests in a physical neighborhood by using a space gravity coefficient when the fuse is not triggered, and generating an atomized continuous execution task package when the condition is met.The application introduces a space gravity field model, solves the dimensional conflict of logical sequencing and physical execution path, and avoids zigzag return scheduling loss.
Owner:CHENGDU ZHIRUILING TECH CO LTD +1

Preparation method of packaging substrate and packaging substrate

The preparation method comprises the steps that a core plate layer is provided, the core plate layer comprises a core plate, the core plate comprises two surfaces which are oppositely arranged, core plate copper layers are arranged on the two surfaces, and the surface of the core plate layer comprises a coding area and a functional area. Codes are printed in the code printing area of the core board layer to form through hole codes, and the through hole codes penetrate through the core board layer. And patterning the core board copper layer of the functional area so as to respectively form core board circuit layers on the two surfaces of the core board. Laminating a layer-adding film on the surface of the circuit layer of the core board, and controlling the layer-adding film to be formed in the functional area far away from the coding area. Forming a circuit layer on the layer-adding film, wherein the layer-adding film and the circuit layer form an adding layer; when the layer-adding film is pressed, the layer-adding film is controlled to be far away from the coding area so as to prevent the layer-adding film from covering the coding area and affecting the readability of the through hole code of the coding area. Therefore, through the scheme of'one code to the bottom ', the yield and the production efficiency of the PCB are improved, and the manufacturing cost is reduced.
Owner:XIAMEN ANJIELI MEIWEI TECH CO LTD +1

Step flow card display method and device

The embodiment of the invention provides a step flow card display method and device and electronic equipment, and belongs to the technical field of data processing, and the method comprises the steps that an original ternary structure set SFC = {CD, SF and OP} is constructed, the CD is a single card data structure, the SF is a card set structure, and the OP is a single card operation mode structure; and constructing the same first DOM tree and second DOM tree in the memory based on the SFC. Obtaining a first change parameter C1 through card state change, obtaining a second change parameter C2 through an optimization function, and respectively inputting the first change parameter C1 and the second change parameter C2 into the two DOM trees to obtain operation times N1 and N2; when (N1-N2) / N1 is greater than a preset value alpha 1, updating the SFC by using C2 to obtain an SFU; and obtaining the actual width W1 of the screen and the display width W2 of the cards in the SFU, calculating alpha 2 when W2 is greater than W1, and generating 1 + round (n * alpha 2) card display according to the alpha 2. According to the method, the smoothness and the stability of card display are remarkably improved, and the response speed of the whole system and the user interaction experience are effectively improved.
Owner:ONE NETWORK INTEROPERABILITY (BEIJING) TECH CO LTD

Long text processing method, device, equipment and medium based on classification model

The present application relates to a long text processing method, device, equipment and medium based on a classification model, wherein the method includes: obtaining long text data to be compressed, and preprocessing the long text data to be compressed to obtain text features and preprocessed text; performing algorithm prediction on text features through a pre-trained classification model to obtain a target compression algorithm combination; performing data encoding on the preprocessed text to obtain an encoded text; performing multi-level compression processing on the encoded text based on the target compression algorithm combination to obtain target compressed data; if a data recovery instruction is received, performing multi-level decompression and data recovery processing on the target compressed data to obtain target recovered data. The present application improves the compression rate of long texts, reduces compression costs, ensures the integrity and readability of compressed texts, and realizes rapid decompression and recovery of compressed data.
Owner:HANGZHOU HUACE FILM & TELEVISION TECH CO LTD

Delivery receipt generation system for delivery service

ActiveCN121997890AAchieve deep integrationKeep visual details intactSemantic analysisAlgorithmGradation
The invention discloses a delivery receipt generation system for delivery service, and relates to the technical field of delivery receipt generation, and the system comprises a data collection module which carries out the original image collection of a physical paper delivery receipt, generates a physical pixel unit and a template pixel unit, and constructs a mixed visual pixel pool; the visual substrate processing module is used for extracting a real object pixel unit from the mixed visual pixel pool, executing gray scale inverse conversion, positioning a putting node based on a gray scale difference value and randomly putting fragments to form a cross-domain visual fragment fusion substrate; the anti-counterfeiting character embedding module is used for extracting a special-shaped embedding carrier of a cross-domain visual fragment fusion substrate based on connected domain analysis, and embedding the fine character units into the corresponding special-shaped embedding carrier to obtain a visual fusion image; and the output module is used for executing secondary gray scale inverse restoration on the non-core anti-counterfeiting region of the visual fusion image to generate an electronic delivery receipt with an anti-counterfeiting traceability mark, so that the beneficial effect that the visual form of the receipt cannot be copied is realized.
Owner:张家港保税数据科技有限公司

Converter station DC secondary control and protection equipment network topological graph generation method and system

The invention discloses a method and system for generating a network topological graph of converter station direct current secondary control and protection equipment, and the method comprises the steps: obtaining the type of a topological graph to be generated of the converter station direct current secondary control and protection equipment, the type of the topological graph comprises a topological overview graph and topological sub-graphs, and the topological sub-graphs comprise an SCADA LAN sub-graph, an STN LAN sub-graph, an FIELD LAN sub-graph, a CONNTROL LAN sub-graph and a TDM LAN sub-graph; based on the to-be-generated topological graph type, analyzing the SPCD file of the converter station to obtain topological data of the to-be-generated topological graph type; drawing is carried out based on the topological data, and a topological graph of the converter station direct current secondary control and protection equipment is generated; according to the method, the SPCD data of the converter station can be analyzed, and the structured topological data is defined, so that the topological graph is automatically generated, and the problem of errors or low precision in manual drawing is avoided.
Owner:NANJING FIVE MINING WISDOM ELECTRIC POWER TECHNOLOGY CO LTD

Material intelligent question-answering and performance prediction method and system based on knowledge graph

The invention discloses a material intelligent question-answering and performance prediction method and system based on a knowledge graph, relates to the technical field of material prediction and analysis, and adopts a new knowledge graph construction mode to carry out structured representation on material data, so that material information is reserved to the greatest extent, and the readability of the graph is ensured. According to the system framework provided by the invention, accurate material knowledge questions and answers can be realized, knowledge completion is realized in a similarity calculation mode so as to complete material performance prediction, and the system framework has important significance in accelerating discovery of new materials and mining potential information of the materials.
Owner:COMP NETWORK INFORMATION CENT CHINESE ACADEMY OF SCI

Two-dimensional code sample screening method and system based on multiple modes and storage medium

The invention discloses a multi-mode-based two-dimensional code sample screening method and system and a storage medium. The method comprises the steps of obtaining an original two-dimensional code picture set; extracting various types of feature information of two-dimensional code pictures contained in the original two-dimensional code picture set to obtain a multi-modal feature value corresponding to each two-dimensional code picture in the original two-dimensional code picture set; and comparing the multi-modal feature value corresponding to each two-dimensional code picture with a set multi-modal feature value, screening out the two-dimensional code pictures meeting a preset condition, and obtaining a two-dimensional code sample set. According to the method provided by the embodiment of the invention, the two-dimensional code pictures in the two-dimensional code picture set can be automatically screened, the screening efficiency is improved, the integrity and readability of the two-dimensional codes in the two-dimensional code sample set for network model training are ensured, and the recognition accuracy of the network model is further improved.
Owner:深圳开鸿数字产业发展有限公司

Road signboard convenient to move

The utility model relates to the related technical field of road signboards, and discloses a road signboard convenient to move, which is characterized in that a damping part is provided with a first fixing plate, a fixing cylinder is fixedly welded on the upper surface of the first fixing plate, and a bolt is rotatably mounted on the outer side surface of the fixing cylinder; a first adjusting rod is fixedly installed on the upper surface of the fixed cylinder, a rotating handle is rotatably installed on the outer side face of the first adjusting rod, an adjusting rod body is fixedly connected to the upper surface of the first adjusting rod, and a flickering lampshade is arranged at the top end of the adjusting rod body; the rotating clamping plate arranged in the adjusting moving box can push the moving wheels to extend downwards and retract into the lifting cavity by rotating the rotating connecting rod, and by means of the effect, the use state of the lifting device can be changed according to self requirements when the lifting device moves outdoors.
Owner:SI BAI RUI IND (SHANGHAI) CO LTD