Product label identification method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of artificial intelligence, can be applied to business system platforms of medical health, financial science and technology and the like, and discloses a product label identification method and device, computer equipment and a storage medium, various sample data related to product label identification are collected, and the sample data comprise text data, picture data and voice data; preprocessing all the sample data to obtain a fine tuning data set, and performing fine tuning on a pre-trained large language model by using the fine tuning data set to generate a product label identification model; obtaining description information of a target product, wherein the description information is generated by utilizing one or more of text information, picture information and voice information related to the target product; formatting the description information to obtain target information, and generating a label identification result of the target product through the product label identification model; therefore, efficient and accurate label identification of the target product can be realized.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a product label recognition method, device, computer equipment, and computer-readable storage medium. Background Art

[0002] Currently, in certain industries, product label recognition is a crucial tool for classifying, retrieving, and analyzing related products. For example, in the insurance industry, as the insurance market continues to evolve and the variety and complexity of insurance products continue to increase, the requirements for product label recognition and management are also becoming increasingly stringent. However, traditional label recognition methods currently rely primarily on manual classification or rule-based systems. These methods suffer from numerous issues with recognition efficiency and accuracy, making them difficult to meet the current needs of the insurance industry.

[0003] At the same time, the healthcare sector generates massive amounts of data, including electronic medical records, medical images, laboratory test results, and patient health monitoring data. The rapid growth of this data makes traditional label recognition and management methods difficult to cope with. Manual classification and rule-based systems are inefficient and prone to errors when processing massive amounts of data.

[0004] Meanwhile, in the field of financial technology, the development of financial technology has driven innovation in financial products, resulting in the emergence of many new financial products, such as digital currencies, blockchain-based financial products, and smart investment advisory services. The structure and characteristics of these products differ significantly from traditional financial products, making it difficult for traditional label identification and management methods to efficiently and accurately classify and label them.

[0005] Based on this, how to provide a product label identification method, device, computer equipment and computer-readable storage medium that can achieve efficient and accurate label identification of target products is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0006] In view of the above-mentioned deficiencies in the prior art, the object of the present invention is to provide a product label recognition method, apparatus, computer equipment and computer-readable storage medium, aiming to solve the problem of how to achieve efficient and accurate label recognition of target products.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] In a first aspect, the present invention provides a product label recognition method, comprising:

[0009] Collecting various sample data related to product label recognition, including text data, image data, and voice data;

[0010] Preprocessing all the sample data to obtain a fine-tuning dataset, and using the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model;

[0011] Obtaining description information of a target product, where the description information is generated using one or more of text information, image information, and voice information related to the target product;

[0012] The description information is formatted to obtain target information, and the target information is used as input to generate a label recognition result of the target product through the product label recognition model.

[0013] In a second aspect, the present invention provides a product label identification device, comprising:

[0014] A collection module, configured to collect various sample data related to product label recognition, including text data, image data, and voice data;

[0015] A fine-tuning module is used to pre-process all the sample data to obtain a fine-tuning dataset, and use the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model;

[0016] An acquisition module, configured to acquire description information of a target product, wherein the description information is generated using one or more of text information, image information, and voice information related to the target product;

[0017] The generation module is used to format the description information to obtain target information, take the target information as input, and generate a label recognition result of the target product through the product label recognition model.

[0018] In a third aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the product label identification method as described above when executing the computer program.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the product label identification method as described above.

[0020] Compared with the prior art, the present invention provides a product label recognition method, apparatus, computer equipment and computer-readable storage medium, wherein a variety of sample data related to product label recognition are collected, the sample data including text data, image data and voice data; all the sample data are preprocessed to obtain a fine-tuning data set, and the pre-trained large language model is fine-tuned using the fine-tuning data set to generate a product label recognition model; descriptive information of the target product is obtained, and the descriptive information is generated using one or more of the text information, image information and voice information related to the target product; the descriptive information is formatted to obtain target information, and the target information is used as input to generate a label recognition result of the target product through the product label recognition model; thereby, the present invention can achieve efficient and accurate label recognition for the target product. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 A schematic diagram of an application environment of a product label recognition method provided by an embodiment of the present invention.

[0023] Figure 2 A flowchart of a product label recognition method provided by one embodiment of the present invention.

[0024] Figure 3 A schematic diagram of a program module of a product label identification device provided by one embodiment of the present invention.

[0025] Figure 4 A schematic diagram of the structure of a computer device provided in one embodiment of the present invention.

[0026] Figure 5 Another structural diagram of a computer device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0027] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0028] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof.

[0029] It will also be understood that the term "and / or" used in the present description and appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0030] As used in the present specification and the appended claims, the term "if" may be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" may be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0031] In addition, in the description of the present specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0032] References to "one embodiment" or "some embodiments" in the present specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present invention. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in yet other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0033] It should be understood that the order of execution of the steps in the following embodiments does not necessarily mean the order in which they are executed. The order in which each process is executed should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0034] In order to illustrate the technical solution of the present invention, specific embodiments are provided below.

[0035] An embodiment of the present invention provides a product label recognition method that can be applied in the following situations: Figure 1In the application environment shown, the client and server communicate via a network. The client includes, but is not limited to, PDAs, desktop computers, laptops, ultra-mobile personal computers (UMPCs), netbooks, cloud computing devices, personal digital assistants (PDAs), and other computer devices. The server can be a standalone server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0036] See also Figure 2 An embodiment of the present invention provides a product label recognition method, wherein the method comprises the following steps:

[0037] S100: Collect various sample data related to product label recognition, wherein the sample data includes text data, image data, and voice data;

[0038] S200: Preprocess all the sample data to obtain a fine-tuning dataset, and use the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model;

[0039] S300: Obtain description information of a target product, where the description information is generated using one or more of text information, image information, and voice information related to the target product;

[0040] S400: Format the description information to obtain target information, use the target information as input, and generate a label recognition result for the target product through the product label recognition model.

[0041] In practice, the product label recognition method of this embodiment achieves efficient and accurate label recognition of target products by combining multimodal data (text, images, and voice) with fine-tuning technology of a pre-trained large language model. The specific analysis is as follows:

[0042] 1. Fusion of multimodal data

[0043] Comprehensive use of text data, image data, and voice data: By collecting a variety of sample data related to product label recognition, including text data, image data, and voice data, the method of this embodiment can comprehensively describe the product from multiple perspectives. Text data can provide detailed product information and user reviews, image data can show the product's appearance and usage scenarios, and voice data can capture users' verbal descriptions and reviews of the product. This fusion of multimodal data enables a more comprehensive understanding of the product's features and attributes during subsequent model fine-tuning, thereby improving the accuracy and comprehensiveness of label recognition.

[0044] 2. Construction of preprocessing and fine-tuning dataset

[0045] Preprocessing step: Preprocessing all sample data can effectively improve the quality of the data and make it more suitable for fine-tuning training of the model.

[0046] Construction of a fine-tuning dataset: The fine-tuning dataset obtained through preprocessing is used to fine-tune the pre-trained large language model. The pre-trained large language model has already learned common language and image features on large-scale data. By fine-tuning on a domain-specific fine-tuning dataset, the generated product label recognition model can be better adapted to the product label recognition task, improving the accuracy and generalization ability of the product label recognition model.

[0047] 3. Generate description information of target products

[0048] Generate multimodal description information: Obtain description information for the target product. This description information can be generated using one or more of the following: text information, image information, and voice information related to the target product. This flexible description information generation method enables the method of this embodiment to adapt to different application scenarios and data sources, improving its versatility and practicality.

[0049] 4. Formatting and model input

[0050] Formatting: Format the target product description information to meet the input requirements of the product label recognition model. Formatting ensures the consistency and standardization of the input content, helping the product label recognition model to process and analyze data more efficiently.

[0051] The model generates label recognition results: The formatted target information is fed into a fine-tuned product label recognition model, which generates accurate label recognition results based on the input target information. Fine-tuning the pre-trained large language model enables the generated product label recognition model to accurately identify the key features and attributes of the target product based on the characteristics of the target information, thereby generating labels that match the product's characteristics.

[0052] 5. Efficiency and accuracy

[0053] Efficiency: By fine-tuning a pre-trained large language model, this method can quickly adapt to new product label recognition tasks, reducing the time and computing resources required to train the model from scratch. Furthermore, the integration of multimodal data enables the product label recognition model to gain a more comprehensive understanding of the product, improving its efficiency.

[0054] Accuracy: The pre-trained large language model learns rich feature representations on large-scale data. Fine-tuning on a domain-specific fine-tuning dataset enables the generated product label recognition model to better adapt to the task, improving its accuracy. The fusion of multimodal data further enhances the product label recognition model's ability to understand and identify product features, making the generated label recognition results more accurate and comprehensive.

[0055] The product label recognition method of this embodiment, through the fusion of multimodal data, the construction of preprocessing and fine-tuning data sets, the generation of descriptive information of the target product, as well as formatting processing and model input, can achieve efficient and accurate label recognition of the target product, and can meet the needs of different industries for product label recognition.

[0056] It is understandable that the product label recognition method provided in the embodiment of the present invention can be applied to product label recognition scenarios related to the insurance industry. The following is a specific example:

[0057] Example: Insurance product label recognition in the insurance industry

[0058] In the insurance industry, as the insurance market continues to develop, the variety and complexity of insurance products continues to increase. Insurance products include, but are not limited to, life insurance, health insurance, property insurance, and auto insurance. Each product has its own unique terms, coverage, and target customer groups. Accurate tag recognition is crucial for the classification, retrieval, risk assessment, and recommendation of insurance products. Traditional tag management methods primarily rely on manual classification or rule-based systems, which are inefficient and prone to errors. Therefore, the product tag recognition method provided in the embodiments of the present invention can effectively address this problem.

[0059] Specific implementation steps

[0060] 1. Collect sample data

[0061] Sample data source:

[0062] Insurance product brochures: Obtain the text data of insurance product brochures from insurance companies. This data contains detailed information about the product, such as insurance terms, coverage, and premium structure.

[0063] Insurance product promotional images: Obtain promotional images of insurance products from insurance companies or insurance agencies. These images can show the key information and features of the products.

[0064] Customer feedback voice: Obtain customer feedback voice data on insurance products from insurance companies or online insurance platforms. This data can include customer descriptions of product satisfaction, claims experience, etc.

[0065] Sample data collection:

[0066] Collect sample data for various insurance products, including text, image, and voice data. For example, collect instruction manuals, promotional images, and customer feedback audio for 100 common insurance products.

[0067] 2. Preprocessing and model fine-tuning

[0068] Preprocessing:

[0069] Text data: Insurance product instructions are processed through word segmentation, stop word removal, part-of-speech tagging, and text vectorization.

[0070] Image data: Insurance product promotional images are cropped, normalized, enhanced, and feature generated.

[0071] Voice data: Perform noise reduction, endpoint detection, parameter extraction, and voice signal vectorization on customer feedback voice.

[0072] Fine-tuning dataset construction: The preprocessed sample data is integrated into a unified dataset as the fine-tuning dataset.

[0073] Model fine-tuning: Use the fine-tuning dataset to fine-tune the pre-trained large language model to generate an insurance product label recognition model.

[0074] 3. Get the description information of the target product

[0075] Target product: Assume that the target product is a new type of health insurance product.

[0076] Description information generation:

[0077] Text information: Obtain the text of the health insurance product instructions from the insurance company.

[0078] Image information: Obtain promotional images of the health insurance product from the insurance company.

[0079] Voice information: Obtain voice feedback from customers on health insurance products from insurance companies.

[0080] Descriptive information integration: Integrate the above text information, image information and voice information into the descriptive information of the target product.

[0081] 4. Formatting and label recognition

[0082] Formatting:

[0083] Format the target product description to meet the input requirements of the insurance product label recognition model. For example, standardize text information, resize and convert images, and proofread and format the transcript of voice information.

[0084] Tag identification:

[0085] The formatted target information is input into the insurance product label recognition model, and the model generates the label recognition result of the health insurance product.

[0086] For example, the generated tag recognition results may include:

[0087] Product name: XX Health Insurance, Product type: Health Insurance, Coverage: Major diseases, Hospitalization, Premium structure: Annual payment, Monthly payment, Target customers: Young and middle-aged groups, Claim features: Quick claims, Online application.

[0088] Application Effect

[0089] The method of the present invention efficiently and accurately identifies key information about target insurance products and generates corresponding tags. These tags can be used for insurance product classification, retrieval, risk assessment, and recommendation, helping insurance companies quickly find products suitable for their customers and improving the efficiency and quality of insurance services. Furthermore, the method of the present invention reduces the errors and inefficiencies associated with manual classification and rule-based systems, thereby improving the overall operational efficiency of the insurance industry.

[0090] Of course, the product label recognition method provided by the embodiment of the present invention can also be applied to product label recognition scenarios related to the medical and health field. The following is a specific example:

[0091] Example: Drug label recognition in healthcare

[0092] In the healthcare sector, drug label recognition is crucial for drug classification, retrieval, and management. Drug labels typically include key information such as the drug name, ingredients, indications, usage and dosage, and adverse reactions. This information is crucial for doctors' prescribing decisions, pharmacists' drug dispensing, and patients' medication guidance. However, traditional drug label management methods primarily rely on manual classification or rule-based systems, which are inefficient and prone to errors. Therefore, the product label recognition method provided in embodiments of the present invention can effectively address this problem.

[0093] Specific implementation steps

[0094] 1. Collect sample data

[0095] Sample data source:

[0096] Drug instructions: Obtain text data of drug instructions from drug manufacturers. This data contains detailed information about the drug, such as ingredients, indications, usage and dosage, etc.

[0097] Drug packaging images: Obtain image data of drug packaging from drug manufacturers or medical institutions. These images can show the appearance of the drug, packaging labels, and other information.

[0098] Patient voice feedback: Obtain patient voice feedback data on drugs from medical institutions or online medical platforms. This data may include patients' descriptions of drug effects, adverse reactions, etc.

[0099] Sample data collection:

[0100] Collect sample data for a variety of medicines, including text, image, and voice data. For example, collect instruction manuals, packaging images, and patient feedback for 100 common medicines.

[0101] 2. Preprocessing and model fine-tuning

[0102] Preprocessing:

[0103] Text data: Perform word segmentation, stop word removal, part-of-speech tagging, and text vectorization on drug instructions.

[0104] Image data: perform cropping, normalization, data enhancement, and feature generation on pharmaceutical packaging images.

[0105] Speech data: The patient feedback speech is processed through noise reduction, endpoint detection, parameter extraction, and speech signal vectorization.

[0106] Fine-tuning dataset construction: The preprocessed sample data is integrated into a unified dataset as the fine-tuning dataset.

[0107] Model fine-tuning: Use the fine-tuning dataset to fine-tune the pre-trained large language model to generate a drug label recognition model.

[0108] 3. Get the description information of the target product

[0109] Target product: Assume that the target product is a new drug for the treatment of hypertension.

[0110] Description information generation:

[0111] Text information: Obtain the drug instructions text from the drug manufacturer.

[0112] Image information: Obtain packaging images of the drug from the drug manufacturer.

[0113] Voice information: Obtain patient voice feedback on the drug from medical institutions.

[0114] Descriptive information integration: Integrate the above text information, image information and voice information into the descriptive information of the target product.

[0115] 4. Formatting and label recognition

[0116] Formatting:

[0117] Format the target product description to meet the input requirements of the drug label recognition model. For example, standardize text, resize and convert images, and proofread and format the transcript of audio.

[0118] Tag identification:

[0119] The formatted target information is input into the drug label recognition model, and the model generates the label recognition result of the drug (target product).

[0120] For example, the generated tag recognition results may include:

[0121] Drug name: XX hypertension treatment drug, ingredients: XX ingredient 1, XX ingredient 2, indications: hypertension, usage and dosage: once a day, one tablet each time, adverse reactions: headache, dizziness, etc.

[0122] Application Effect

[0123] The method of the present invention efficiently and accurately identifies key information about target drugs and generates corresponding labels. These labels can be used for drug classification, retrieval, and management, helping doctors quickly find the right medication for their patients and improving medical efficiency and quality. Furthermore, the method of the present invention reduces the errors and inefficiencies associated with manual classification and rule-based systems, thereby improving overall operational efficiency in the healthcare sector.

[0124] Of course, the product label recognition method provided by the embodiment of the present invention can also be applied to product label recognition scenarios related to the financial technology field. The following is a specific example:

[0125] Example: Financial product label recognition in the FinTech sector

[0126] In the field of FinTech, the variety and complexity of financial products continue to increase, including digital currencies, blockchain-based financial products, and robo-advisory services. The structure and characteristics of these products differ significantly from traditional financial products, making it difficult for traditional tag recognition and management methods to efficiently and accurately classify and label them. Accurate tag recognition is crucial for the classification, retrieval, risk assessment, and recommendation of financial products. Therefore, the product tag recognition method provided by the embodiments of the present invention can effectively address this problem.

[0127] Specific implementation steps

[0128] 1. Collect sample data

[0129] Sample data source:

[0130] Financial product instructions: Obtain the text data of financial product instructions from financial institutions. This data contains detailed information about the product, such as product type, risk level, and return characteristics.

[0131] Product promotional images: Obtain promotional images of financial products from financial institutions or fintech platforms. These images can showcase the key information and features of the products.

[0132] Customer feedback voice: Obtain customer feedback voice data on financial products from financial institutions or online financial platforms. This data may include descriptions of customer satisfaction with the product, risk perception, etc.

[0133] Sample data collection:

[0134] Collect sample data for various financial products, including text, image, and voice data. For example, collect instruction manuals, promotional images, and customer feedback audio for 100 common financial products.

[0135] 2. Preprocessing and model fine-tuning

[0136] Preprocessing:

[0137] Text data: Perform word segmentation, stop word removal, part-of-speech tagging, and text vectorization on financial product instructions.

[0138] Image data: Product promotion images are cropped, normalized, enhanced, and feature-generated.

[0139] Voice data: Perform noise reduction, endpoint detection, parameter extraction, and voice signal vectorization on customer feedback voice.

[0140] Fine-tuning dataset construction: The preprocessed sample data is integrated into a unified dataset as the fine-tuning dataset.

[0141] Model fine-tuning: Use the fine-tuning dataset to fine-tune the pre-trained large language model to generate a financial product label recognition model.

[0142] 3. Get the description information of the target product

[0143] Target product: Assume that the target product is a new type of smart investment advisory service.

[0144] Description information generation:

[0145] Text information: Obtain the text of the instruction manual of the smart investment advisory service from the financial institution.

[0146] Image information: Obtain promotional images of the smart investment advisory service from financial institutions.

[0147] Voice information: Obtain voice feedback from financial institutions regarding the robo-advisory service.

[0148] Descriptive information integration: Integrate the above text information, image information and voice information into the descriptive information of the target product.

[0149] 4. Formatting and label recognition

[0150] Formatting:

[0151] Format the target product's description to meet the input requirements of the financial product label recognition model. For example, standardize text, resize and convert images, and proofread and format the transcripts of voice messages.

[0152] Tag identification:

[0153] The formatted target information is input into the financial product label recognition model, and the model generates the label recognition result of the smart investment advisory service (target product).

[0154] For example, the generated tag recognition results may include:

[0155] Product name: XX smart investment advisory service, product type: smart investment advisor, risk level: medium-low risk, return characteristics: steady growth, service features: personalized investment portfolio, real-time risk monitoring, and intelligent recommendations.

[0156] Application Effect

[0157] The method of the present invention efficiently and accurately identifies key information about target financial products and generates corresponding tags. These tags can be used for classification, retrieval, risk assessment, and recommendation of financial products, helping financial institutions quickly find products suitable for their customers and improving the efficiency and quality of financial services. Furthermore, the method of the present invention reduces the errors and inefficiencies associated with manual classification and rule-based systems, thereby improving overall operational efficiency in the financial technology sector.

[0158] Furthermore, in one embodiment, the product label recognition method, wherein the collecting of a variety of sample data related to product label recognition, specifically includes:

[0159] Identify various sample data sources related to product label identification;

[0160] Based on the sample data source, collecting various sample data;

[0161] Conducting preliminary screening of the collected sample data to remove duplicate or invalid data;

[0162] Perform label verification on the sample data after preliminary screening to ensure that each sample data contains a corresponding labeled label;

[0163] The verified sample data are classified and stored.

[0164] During specific implementation, the specific implementation process of the steps in this embodiment is roughly as follows:

[0165] 1. Determine the source of sample data:

[0166] Analyze the type and characteristics of the product and determine the type of data that needs to be collected, such as text, pictures, voice, etc.

[0167] Determine the source of sample data, for example:

[0168] Text data: product manuals, user reviews, online forums, e-commerce platforms, etc.

[0169] Image data: product packaging, usage scenarios, promotional pictures, etc.

[0170] Voice data: user feedback recordings, customer service call recordings, etc.

[0171] Record detailed information about the source of each sample data, including source channel, data format, acquisition method, etc.

[0172] 2. Collect sample data:

[0173] Design data collection tools or scripts to collect data from different sources automatically or manually.

[0174] For text data, web crawler technology is used to obtain text content from web pages, forums, etc.

[0175] For image data, relevant images are taken by image acquisition equipment or downloaded from the Internet.

[0176] For voice data, use voice capture equipment to record user feedback or customer service calls.

[0177] Ensure that the collected data is diverse and representative, covering different product types and user scenarios.

[0178] 3. Preliminary screening of sample data:

[0179] Use data cleaning tools or scripts to perform preliminary checks on the collected data.

[0180] For text data, use text similarity algorithms (such as cosine similarity) to detect and remove duplicate content.

[0181] For image data, an image hash algorithm (such as perceptual hashing algorithm) is used to detect and remove duplicate images.

[0182] For voice data, voice activity detection (VAD) technology is used to remove silent segments to ensure the validity of voice data.

[0183] Delete obviously invalid data, such as blank text, blurry images, and speech with excessive noise.

[0184] 4. Label verification:

[0185] Design a label verification process to ensure that each sample data has a corresponding label.

[0186] For text data, the text content is manually checked to ensure that the labels accurately reflect the core information of the text.

[0187] For image data, key features and objects in the images are manually annotated to ensure accurate labels.

[0188] For voice data, the voice content is manually transcribed to ensure that the label accurately reflects the core information of the voice.

[0189] Use automated tools to assist with label verification. For example, preliminarily label the data using a pre-trained classification model, followed by manual review and correction.

[0190] 5. Classify and store sample data:

[0191] Design data storage architecture and classify storage according to data type and tag information.

[0192] Use a database or file system to store sample data, for example:

[0193] Text data: stored as structured database tables or text files.

[0194] Image data: stored as image files, organized in folder structures by category and label.

[0195] Voice data: stored as audio files, organized in folder structures by category and tag.

[0196] Add metadata to each sample data to record data source, collection time, label information, etc.

[0197] Ensure the security and accessibility of data storage, set up appropriate access permissions and backup mechanisms.

[0198] Through the above steps, sample data related to product label recognition can be efficiently collected, screened, verified and stored, providing high-quality data support for subsequent model training and label recognition.

[0199] Furthermore, in one embodiment, the product label recognition method, wherein the preprocessing of all the sample data to obtain a fine-tuning dataset, and the use of the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model, specifically includes:

[0200] Preprocessing all the sample data, including word segmentation, stop word removal, part-of-speech tagging, and text vectorization on the text data; image cropping, normalization, data enhancement, and feature generation on the image data; and noise reduction, endpoint detection, parameter extraction, and speech signal vectorization on the speech data;

[0201] Performing feature fusion on the preprocessed text data, the image data, and the voice data to form a unified feature representation;

[0202] Divide the fused feature data into training set, validation set and test set;

[0203] Fine-tuning the pre-trained large language model using the training set, and adjusting parameters of the large language model to adapt to the product label recognition task;

[0204] The fine-tuned large language model is trained and tested using the validation set and the test set. When the test results meet the preset model performance requirements, the product label recognition model is generated.

[0205] During specific implementation, the specific implementation process of the steps in this embodiment is roughly as follows:

[0206] 1. Preprocess sample data

[0207] Step description: Preprocess all sample data to ensure unified data format and reliable quality.

[0208] Text data:

[0209] Word segmentation: Split the text data into words or phrases for subsequent analysis.

[0210] Stop word removal: Delete common meaningless words such as "de" (of), "shi" (is), etc., reduce the data volume, and improve processing efficiency.

[0211] Part-of-speech tagging: Tag the part of speech for each word, such as noun, verb, etc., to enhance semantic understanding.

[0212] Text vectorization: Convert the text into a numerical vector for model processing. Common methods include TFIDF, Word2Vec, etc.

[0213] Image data:

[0214] Image cropping: Remove irrelevant parts of the image and focus on the key area.

[0215] Normalization: Adjust the size and pixel value range of the image to meet the model input requirements.

[0216] Data augmentation: Increase data diversity through operations such as rotation, flipping, scaling, etc., to improve the model generalization ability.

[0217] Feature generation: Extract key features of the image, such as edges, textures, etc., to provide more useful information for the model.

[0218] Voice data:

[0219] Noise reduction: Remove background noise in the voice signal to improve voice quality.

[0220] Endpoint detection: Determine the start and end points of the voice signal and remove the silent parts.

[0221] Parameter extraction: Extract key parameters of the voice signal, such as MFCC, fundamental frequency, etc., to characterize the physical properties of the voice.

[0222] Voice signal vectorization: Convert the voice signal into a numerical vector for model processing.

[0223] 2. Feature fusion

[0224] Step description: Perform feature fusion on the preprocessed text data, image data, and voice data to form a unified feature representation.

[0225] Feature extraction: Extract key features from each data type, such as the TFIDF vector of the text, the feature vector of the image, the MFCC vector of the voice, etc.

[0226] Feature alignment: Align feature vectors of different data types to the same dimension to ensure that they can be processed together.

[0227] Feature fusion: Use fusion techniques (such as weighted summation, splicing, etc.) to merge feature vectors of different data types into a unified feature vector to provide comprehensive input information for the model.

[0228] 3. Dataset Division

[0229] Step description: Divide the fused feature data into training set, validation set and test set.

[0230] Division ratio: Determine the ratio of training set, validation set and test set based on the amount of data and task requirements, such as 70% training set, 15% validation set, and 15% test set.

[0231] Random partitioning: Use random sampling to divide the data into three sets to ensure that the data in each set is evenly distributed.

[0232] Data preservation: The divided data sets are stored separately to facilitate subsequent model training and testing.

[0233] 4. Model fine-tuning

[0234] Step Description: Use the training set to fine-tune the pre-trained large language model and adjust the model parameters to adapt to the product label recognition task.

[0235] Load a pre-trained model: Select a suitable large language model (such as BERT, GPT, etc.) and load its pre-trained parameters.

[0236] Fine-tuning settings: Adjust the model's training parameters, such as learning rate and batch size, according to task requirements.

[0237] Training process: Use the training set data to train the model, adjust the model parameters through back propagation, and optimize the model performance.

[0238] Validation process: During the training process, the validation set is regularly used to evaluate the model performance, prevent overfitting, and adjust the training strategy.

[0239] 5. Model Evaluation and Generation

[0240] Step description: Use the validation set and test set to evaluate the fine-tuned model. When the test results meet the preset model performance requirements, generate a product label recognition model.

[0241] Performance evaluation: Use the validation set and test set to evaluate the model and calculate performance indicators (such as accuracy, recall, F1 value, etc.).

[0242] Performance requirements: Set minimum requirements for model performance, such as accuracy must reach above 90%.

[0243] Model selection: Based on the evaluation results, the model with the best performance is selected as the product label recognition model.

[0244] Model saving: Save the final generated model parameters to facilitate subsequent product label recognition tasks.

[0245] Through the above steps, sample data can be efficiently preprocessed, features fused, dataset partitioned, model fine-tuned, and evaluated, ultimately generating a high-quality model suitable for product label recognition tasks.

[0246] Furthermore, in one embodiment, the product label recognition method, wherein obtaining the description information of the target product specifically includes:

[0247] Acquire one or more of the text information, the image information, and the voice information related to the target product input by the target user;

[0248] Preliminarily classifying the content input by the target user, determining the type of the input content, and performing preliminary processing based on the type;

[0249] The preliminarily processed content is parsed to extract characteristic information of the target product, and the characteristic information is integrated to generate the description information of the target product.

[0250] During specific implementation, the specific implementation process of the steps in this embodiment is roughly as follows:

[0251] 1. Get user input

[0252] Collect text, pictures or voice information related to the target product entered by the user.

[0253] 2. Preliminary classification and processing

[0254] Classify input according to type (text, image, voice) and perform preliminary processing, such as text formatting, image cropping, and voice noise reduction.

[0255] 3. Parsing and feature extraction

[0256] After parsing the processed content, extract the characteristic information of the target product, such as keywords in the text, object features in the image, and key sentences in the speech.

[0257] 4. Integrate and generate description information

[0258] Integrate the extracted feature information to generate descriptive information of the target product, ensuring that the information is complete and concise.

[0259] Through these steps, the description information of the target product can be efficiently extracted and generated from the user input, providing accurate input for subsequent product label recognition.

[0260] Furthermore, in one embodiment, the product label recognition method, wherein the parsing of the preliminarily processed content, extracting characteristic information of the target product, and integrating the characteristic information to generate the description information of the target product, specifically includes:

[0261] Performing semantic analysis, image analysis, or voice analysis on the preliminarily processed content to extract characteristic information of the target product;

[0262] Standardize the extracted text features, image features or voice features;

[0263] All standardized features are integrated to form the description information of the target product.

[0264] During specific implementation, the specific implementation process of the steps in this embodiment is roughly as follows:

[0265] 1. Content Analysis

[0266] Perform semantic analysis, image analysis, or voice analysis on the initially processed content to extract feature information of the target product.

[0267] Semantic analysis: Perform natural language processing on text content to extract features such as keywords, phrases, entities, and sentiment. Use pre-trained language models (such as BERT) to extract semantic vectors for the text.

[0268] Image Analysis: Perform computer vision analysis on image content to extract features such as objects, scenes, colors, and textures. Use pre-trained image recognition models (such as ResNet) to extract feature vectors from images.

[0269] Speech analysis: Perform speech recognition on the speech content and extract the text information from the speech. Use speech feature extraction technology (such as MFCC) to extract the feature vector of the speech.

[0270] 2. Feature Standardization

[0271] Normalize the extracted text features, image features or voice features.

[0272] Text feature normalization: Normalize the text feature vector to ensure that its value is between 0 and 1.

[0273] Image feature normalization: Normalize the image feature vector to ensure that its value is between 0 and 1.

[0274] Speech feature normalization: Normalize the speech feature vector to ensure that its value is between 0 and 1.

[0275] 3. Feature Integration

[0276] Integrate all standardized features to form the description information of the target product.

[0277] Feature fusion: Use feature fusion techniques (such as weighted summation, splicing, etc.) to merge feature vectors of different modalities into a unified feature vector.

[0278] Description generation: Generate description information for the target product based on the integrated feature vectors. Natural language generation technology can be used to convert feature vectors into natural language descriptions.

[0279] Through the above steps, the description information of the target product can be efficiently extracted and generated from the user input, providing accurate input for subsequent product label recognition.

[0280] Furthermore, in one embodiment, the product label recognition method, wherein the formatting of the description information to obtain target information, taking the target information as input, and generating a label recognition result of the target product through the product label recognition model, specifically includes:

[0281] Formatting the description information according to the input requirements of the product label recognition model to obtain the target information;

[0282] Inputting the target information into the product label recognition model to generate the label recognition result of the target product;

[0283] Comparing the tag recognition result with a preset standard tag to verify the accuracy of the tag recognition result, and optimizing the tag recognition result based on the comparison result;

[0284] The optimized tag recognition result is output or displayed.

[0285] During specific implementation, the specific implementation process of the steps in this embodiment is roughly as follows:

[0286] 1. Formatting

[0287] Format the description information according to the model input requirements, including text format adjustment, data structure adjustment and normalization processing.

[0288] 2. Input model to generate results

[0289] The formatted target information is input into the product label recognition model to generate the label recognition result.

[0290] 3. Comparative verification and optimization

[0291] The generated tag recognition results are compared with the preset standard tags to verify the accuracy, and the tag recognition results are optimized based on the comparison results.

[0292] 4. Output or display results

[0293] The optimized tag recognition results are output or displayed to the user to ensure the accuracy and readability of the results.

[0294] Through these steps, the label recognition results of the target product can be efficiently generated and optimized to ensure their accuracy and practicality.

[0295] Furthermore, in one embodiment, the product label recognition method, after outputting or displaying the optimized label recognition result, further specifically includes:

[0296] Determine the category to which the target product belongs based on the optimized tag recognition result and a preset product classification strategy;

[0297] Comparing the optimized label recognition results with the preset classification standards, and filing the target product into the corresponding category according to the comparison results;

[0298] The classification results of the target products are stored in a target database.

[0299] During specific implementation, the specific implementation process of the steps in this embodiment is roughly as follows:

[0300] 1. Determine the category of your target product

[0301] Step Description: Determine the category to which the target product belongs based on the optimized tag recognition results and the preset product classification strategy.

[0302] Specific operations:

[0303] Load classification strategy: Load preset product classification strategies, which define the characteristics and labels of different categories.

[0304] Matching labels and categories: Match the optimized label recognition results with the features in the classification strategy to determine the category to which the target product belongs.

[0305] Category determination: Based on the matching results, determine the specific category to which the target product belongs.

[0306] 2. Archive target products

[0307] Step description: Compare the optimized label recognition results with the preset classification standards, and file the target products into the corresponding categories based on the comparison results.

[0308] Specific operations:

[0309] Load Classification Standards: Loads preset classification standards that define specific requirements and conditions for each category.

[0310] Compare labels with standards: Compare the optimized label recognition results with the classification standards to verify whether the labels meet the category requirements.

[0311] Filing operation: Based on the comparison results, the target products are filed into the corresponding categories to ensure the accuracy and consistency of the classification.

[0312] 3. Store classification results

[0313] Step Description: Store the classification results of the target product in the target database.

[0314] Specific operations:

[0315] Connect to database: Establish a connection with the target database to ensure the security and reliability of data storage.

[0316] Data storage: The classification results of the target products are stored in the database, including description information, label recognition results and categories.

[0317] Data indexing: Create indexes for stored data to facilitate subsequent query and analysis.

[0318] Through the above steps, the target products can be efficiently filed into the correct categories, and the classification results can be stored in the target database to ensure the accuracy and security of the data.

[0319] As can be seen from the above method embodiments, the product label recognition method provided by the present invention includes: collecting a variety of sample data related to product label recognition, the sample data including text data, image data and voice data; pre-processing all of the sample data to obtain a fine-tuning data set, using the fine-tuning data set to fine-tune the pre-trained large language model to generate a product label recognition model; obtaining description information of the target product, the description information being generated using one or more of the text information, image information and voice information related to the target product; formatting the description information to obtain target information, using the target information as input, and generating the label recognition result of the target product through the product label recognition model. In this way, the method of the present invention can achieve efficient and accurate label recognition for the target product.

[0320] It should be understood that although the present application provides method operation steps as described in the embodiments or flowcharts, more or fewer operation steps may be included based on conventional or non-creative work, and these operation steps are not necessarily performed in the order of the embodiments or flowcharts. The order of steps listed in the embodiments or flowcharts is only one way of executing the steps among many steps and does not represent the only execution order. It should be noted that there is not necessarily a certain order between the above steps. Those of ordinary skill in the art can understand from the description of the embodiments of the present invention that in different embodiments, the above steps may have different execution orders, that is, they may be executed in parallel, or they may be executed in an interchangeable manner, etc. Moreover, at least a portion of the steps in the embodiments or flowcharts may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. The execution order of these sub-steps or stages is not necessarily to be performed in sequence, but may be executed in turn, alternately or synchronously with other steps or at least a portion of the sub-steps or stages of other steps.

[0321] Based on the above method embodiment, please refer to Figure 3 Another embodiment of the present invention further provides a product label identification device, wherein the device includes:

[0322] A collection module 11 is used to collect various sample data related to product label identification, wherein the sample data includes text data, image data, and voice data;

[0323] A fine-tuning module 12 is configured to pre-process all the sample data to obtain a fine-tuning dataset, and use the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model;

[0324] An acquisition module 13 is configured to acquire description information of a target product, wherein the description information is generated using one or more of text information, image information, and voice information related to the target product;

[0325] The generating module 14 is configured to format the description information to obtain target information, and use the target information as input to generate a label recognition result of the target product through the product label recognition model.

[0326] Furthermore, in one embodiment, the product label recognition device, wherein the collecting of a variety of sample data related to product label recognition specifically includes:

[0327] Identify various sample data sources related to product label identification;

[0328] Based on the sample data source, collecting various sample data;

[0329] Conducting preliminary screening of the collected sample data to remove duplicate or invalid data;

[0330] Perform label verification on the sample data after preliminary screening to ensure that each sample data contains a corresponding labeled label;

[0331] The verified sample data are classified and stored.

[0332] Furthermore, in one embodiment, the product label recognition device, wherein the preprocessing of all the sample data to obtain a fine-tuning dataset, and using the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model, specifically includes:

[0333] Preprocessing all the sample data, including word segmentation, stop word removal, part-of-speech tagging, and text vectorization on the text data; image cropping, normalization, data enhancement, and feature generation on the image data; and noise reduction, endpoint detection, parameter extraction, and speech signal vectorization on the speech data;

[0334] Performing feature fusion on the preprocessed text data, the image data, and the voice data to form a unified feature representation;

[0335] Divide the fused feature data into training set, validation set and test set;

[0336] Fine-tuning the pre-trained large language model using the training set, and adjusting parameters of the large language model to adapt to the product label recognition task;

[0337] The fine-tuned large language model is trained and tested using the validation set and the test set. When the test results meet the preset model performance requirements, the product label recognition model is generated.

[0338] Furthermore, in one embodiment, the product label recognition device, wherein the step of obtaining the description information of the target product specifically includes:

[0339] Acquire one or more of the text information, the image information, and the voice information related to the target product input by the target user;

[0340] Preliminarily classifying the content input by the target user, determining the type of the input content, and performing preliminary processing based on the type;

[0341] The preliminarily processed content is parsed to extract characteristic information of the target product, and the characteristic information is integrated to generate the description information of the target product.

[0342] Furthermore, in one embodiment, the product label recognition device, wherein the parsing of the preliminarily processed content, extracting characteristic information of the target product, and integrating the characteristic information to generate the description information of the target product, specifically includes:

[0343] Performing semantic analysis, image analysis, or voice analysis on the preliminarily processed content to extract characteristic information of the target product;

[0344] Standardize the extracted text features, image features or voice features;

[0345] All standardized features are integrated to form the description information of the target product.

[0346] Furthermore, in one embodiment, the product label recognition device, wherein the step of formatting the description information to obtain target information, taking the target information as input, and generating a label recognition result of the target product through the product label recognition model specifically includes:

[0347] Formatting the description information according to the input requirements of the product label recognition model to obtain the target information;

[0348] Inputting the target information into the product label recognition model to generate the label recognition result of the target product;

[0349] Comparing the tag recognition result with a preset standard tag to verify the accuracy of the tag recognition result, and optimizing the tag recognition result based on the comparison result;

[0350] The optimized tag recognition result is output or displayed.

[0351] Furthermore, in one embodiment, the product label recognition device, after outputting or displaying the optimized label recognition result, further specifically includes:

[0352] Determine the category to which the target product belongs based on the optimized tag recognition result and a preset product classification strategy;

[0353] Comparing the optimized label recognition results with the preset classification standards, and filing the target product into the corresponding category according to the comparison results;

[0354] The classification results of the target products are stored in a target database.

[0355] It should be noted that, in the embodiment of the device of the present invention, the information interaction, execution process and other contents between the above modules are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the aforementioned method embodiment part and will not be repeated here.

[0356] Based on the above method embodiment, another embodiment of the present invention further provides a computer device, which can be a server, and its internal structure diagram can be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the functions or steps on the service side of the product label identification method in any of the above method embodiments are implemented.

[0357] Based on the above method embodiment, another embodiment of the present invention further provides a computer device, which can be a client, and its internal structure diagram can be as follows: Figure 5 As shown. The computer device includes a processor, memory, network interface, display screen, and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, the functions or steps of the client side of the product label identification method in any of the above-mentioned method embodiments are implemented.

[0358] Those skilled in the art will understand that Figure 4 and Figure 5 The structural diagram shown in the figure is only a schematic diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more components than shown in the figure, or combine certain components, or have a different component arrangement.

[0359] The processor referred to herein may be a CPU, other general-purpose processors, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or any conventional processor, etc.

[0360] The memory includes a readable storage medium, an internal memory, etc., wherein the internal memory can be the memory of a computer device, and the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the readable storage medium. The readable storage medium can be a hard disk of the computer device, and in other embodiments, it can also be an external storage device of the computer device, for example, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the computer device. Furthermore, the memory can also include both an internal storage unit of the computer device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of a computer program. The memory can also be used to temporarily store data that has been output or is about to be output.

[0361] Based on the above method embodiments, another embodiment of the present invention further provides a computer-readable storage medium storing a computer program. When executed by a processor, the computer program implements the product label identification method described in any of the above method embodiments. The computer-readable storage medium may be non-volatile or volatile.

[0362] It should be noted that the above-mentioned functions or steps that can be implemented by computer-readable storage media or computer devices, and the technical effects brought about by the functions / steps, can be found in the relevant descriptions in the aforementioned method embodiments. To avoid repetition, they will not be described one by one here.

[0363] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM). The disclosed memory components or memories of the operating environments described herein are intended to comprise one or more of these and / or any other suitable types of memory.

[0364] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, in the embodiment of the device of the present invention, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual application, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other and are not used to limit the scope of protection of the present invention. The specific working process of the units and modules in the above-mentioned device can refer to the corresponding process in the above-mentioned method embodiment, which will not be repeated here. If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium.

[0365] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0366] In the embodiments provided by the present invention, it should be understood that the disclosed apparatus / computer equipment and methods can be implemented in other ways. For example, the apparatus / computer equipment embodiments described above are merely illustrative. For example, the division of modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0367] Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0368] It should be noted that if software tools or components other than those of the Company appear in the embodiments of this application, they are merely for illustration and do not represent actual use. The above embodiments are intended only to illustrate the technical solutions of the present invention, not to limit them. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some of the technical features therein with equivalents. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A product label identification method, characterized in that: include: Collecting various sample data related to product label recognition, including text data, image data, and voice data; Preprocessing all the sample data to obtain a fine-tuning dataset, and using the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model; Obtaining description information of a target product, where the description information is generated using one or more of text information, image information, and voice information related to the target product; The description information is formatted to obtain target information, and the target information is used as input to generate a label recognition result of the target product through the product label recognition model.

2. The product label identification method according to claim 1, characterized in that: The collected product label identification related sample data includes: Identify various sample data sources related to product label identification; Based on the sample data source, collecting various sample data; Conducting preliminary screening of the collected sample data to remove duplicate or invalid data; Perform label verification on the sample data after preliminary screening to ensure that each sample data contains a corresponding labeled label; The verified sample data are classified and stored.

3. The product label identification method according to claim 1, characterized in that: The preprocessing of all the sample data to obtain a fine-tuning dataset, and using the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model, includes: Preprocessing all the sample data, including word segmentation, stop word removal, part-of-speech tagging, and text vectorization on the text data; image cropping, normalization, data enhancement, and feature generation on the image data; and noise reduction, endpoint detection, parameter extraction, and speech signal vectorization on the speech data; Performing feature fusion on the preprocessed text data, the image data, and the voice data to form a unified feature representation; Divide the fused feature data into training set, validation set and test set; Fine-tuning the pre-trained large language model using the training set, and adjusting parameters of the large language model to adapt to the product label recognition task; The fine-tuned large language model is trained and tested using the validation set and the test set. When the test results meet the preset model performance requirements, the product label recognition model is generated.

4. The product label identification method according to claim 1, characterized in that: The step of obtaining the description information of the target product includes: Acquire one or more of the text information, the image information, and the voice information related to the target product input by the target user; Preliminarily classifying the content input by the target user, determining the type of the input content, and performing preliminary processing based on the type; The preliminarily processed content is parsed to extract characteristic information of the target product, and the characteristic information is integrated to generate the description information of the target product.

5. The product label identification method according to claim 4, characterized in that: The step of parsing the preliminarily processed content, extracting characteristic information of the target product, and integrating the characteristic information to generate the description information of the target product includes: Performing semantic analysis, image analysis, or voice analysis on the preliminarily processed content to extract characteristic information of the target product; Standardize the extracted text features, image features or voice features; All standardized features are integrated to form the description information of the target product.

6. The product label identification method according to claim 1, characterized in that: The step of formatting the description information to obtain target information, taking the target information as input, and generating a label recognition result of the target product through the product label recognition model includes: Formatting the description information according to the input requirements of the product label recognition model to obtain the target information; Inputting the target information into the product label recognition model to generate the label recognition result of the target product; Comparing the tag recognition result with a preset standard tag to verify the accuracy of the tag recognition result, and optimizing the tag recognition result based on the comparison result; The optimized tag recognition result is output or displayed.

7. The product label identification method according to claim 6, characterized in that: After outputting or displaying the optimized tag recognition result, the method further includes: Determine the category to which the target product belongs based on the optimized tag recognition result and a preset product classification strategy; Comparing the optimized label recognition results with the preset classification standards, and filing the target product into the corresponding category according to the comparison results; The classification results of the target products are stored in a target database.

8. A product label identification device, characterized in that: include: A collection module, configured to collect various sample data related to product label recognition, including text data, image data, and voice data; A fine-tuning module is used to pre-process all the sample data to obtain a fine-tuning dataset, and use the fine-tuning dataset to fine-tune the pre-trained large language model to generate a product label recognition model; An acquisition module, configured to acquire description information of a target product, wherein the description information is generated using one or more of text information, image information, and voice information related to the target product; The generation module is used to format the description information to obtain target information, take the target information as input, and generate a label recognition result of the target product through the product label recognition model.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the product label identification method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the product label identification method according to any one of claims 1 to 7 is implemented.

Images