Live broadcast e-commerce intelligent quality inspection system and method, electronic equipment and medium

Abstract

The embodiment of the invention provides a live broadcast e-commerce intelligent quality inspection system and method, electronic equipment and a medium, and the system comprises a multi-modal data collection module which is used for collecting live broadcast information in a live broadcast process; the multi-modal feature extraction and fusion module is used for extracting features of the live broadcast information, generating fusion feature representation of the live broadcast information through a multi-modal fusion algorithm, and generating a multi-modal fusion vector; the intelligent quality inspection judgment module is used for analyzing the multi-modal fusion vector through a trained deep learning model and outputting a quality inspection evaluation result of the live broadcast product; the block chain uplink module is used for uplink of the quality inspection evaluation result and the detection information, and non-tampering block chain traceability information is generated on a block chain; and the visual output module is used for displaying the quality inspection evaluation result and the block chain traceability information.

Description

Technical Field

[0001] This document relates to the field of artificial intelligence technology, and in particular to an intelligent quality inspection system, method, electronic device and medium for live e-commerce. Background Technology

[0002] With the rapid development of live-streaming e-commerce for agricultural products, traditional sales methods are gradually being replaced by emerging formats such as live-streaming sales. Live-streaming e-commerce, with its advantages of strong interactivity, intuitive display, and high sales efficiency, is widely welcomed by farmers, e-commerce platforms, and consumers. However, the following problems still exist in the actual operation of live-streaming e-commerce:

[0003] (1) Quality inspection methods are outdated and rely heavily on manual judgment, making it difficult to achieve standardization;

[0004] (2) There is a lack of effective quality certification mechanism during the live broadcast, resulting in low consumer trust.

[0005] (3) The after-sales rights protection mechanism is imperfect and lacks traceable data support;

[0006] (4) Quality inspection data is easily tampered with and lacks a reliable quality certification mechanism.

[0007] While existing technologies have proposed using AI methods such as image recognition for agricultural product quality inspection, they are mostly limited to a single modality and cannot handle the real-time fusion and comprehensive judgment of information such as images, voice, and text in live streaming scenarios. Furthermore, existing quality inspection data has not been integrated with blockchain, failing to achieve reliable recording and traceability of the entire quality inspection process. Therefore, there is an urgent need to build an intelligent quality inspection system that integrates multimodal AI and blockchain. Summary of the Invention

[0008] The purpose of this invention is to provide an intelligent quality inspection system, method, electronic device and medium for live e-commerce, in order to solve the above-mentioned problems in the prior art.

[0009] This invention provides an intelligent quality inspection system for live-streaming e-commerce, comprising:

[0010] A multimodal data acquisition module is used to collect live streaming information during the live broadcast.

[0011] The multimodal feature extraction and fusion module is used to extract features from the live broadcast information and generate a fused feature representation of the live broadcast information through a multimodal fusion algorithm, thereby generating a multimodal fusion vector.

[0012] The intelligent quality inspection and judgment module is used to analyze the multimodal fusion vector through a trained deep learning model and output the quality inspection and evaluation results of the live broadcast product.

[0013] The blockchain on-chain module is used to upload the quality inspection and evaluation results and testing information to the blockchain, generating tamper-proof blockchain traceability information on the blockchain.

[0014] The visualization output module is used to display the quality inspection and evaluation results and blockchain traceability information.

[0015] This invention provides an intelligent quality inspection method for live-streaming e-commerce, used in the aforementioned intelligent quality inspection system for live-streaming e-commerce, comprising:

[0016] Live streaming information is collected during the live broadcast using a multimodal data acquisition module;

[0017] The features of the live broadcast information are extracted by the multimodal feature extraction and fusion module, and the fused feature representation of the live broadcast information is generated by the multimodal fusion algorithm, thus generating a multimodal fusion vector.

[0018] The intelligent quality inspection and judgment module uses a trained deep learning model to analyze the multimodal fusion vector and outputs the quality inspection and evaluation results of the live streaming product.

[0019] The quality inspection and evaluation results and testing information are uploaded to the blockchain through the blockchain on-chain module, generating tamper-proof blockchain traceability information on the blockchain.

[0020] The quality inspection and assessment results and blockchain traceability information are displayed through a visualization output module.

[0021] This invention also provides an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the above-described intelligent quality inspection method for live-streaming e-commerce.

[0022] This invention also provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor, implements the steps of the above-described intelligent quality inspection method for live-streaming e-commerce.

[0023] By employing embodiments of the present invention, real-time, objective, and standardized testing of agricultural product quality in live-streaming e-commerce scenarios is achieved, and blockchain technology is used to ensure the authenticity and immutability of quality inspection information. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in one or more embodiments of this specification or in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a schematic diagram of the intelligent quality inspection system for live-streaming e-commerce according to an embodiment of the present invention;

[0026] Figure 2 This is a detailed structural diagram of the intelligent quality inspection system for live-streaming e-commerce according to an embodiment of the present invention;

[0027] Figure 3 This is a flowchart of the multimodal data acquisition and fusion process according to an embodiment of the present invention;

[0028] Figure 4 This is a flowchart of the quality inspection and evaluation and blockchain on-chain process according to an embodiment of the present invention;

[0029] Figure 5 This is a flowchart of the intelligent quality inspection method for live-streaming e-commerce according to an embodiment of the present invention;

[0030] Figure 6 This is a schematic diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0031] To enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the technical solutions in one or more embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, and not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of this document.

[0032] System Implementation Examples

[0033] According to an embodiment of the present invention, a smart quality inspection system for live-streaming e-commerce is provided, which is used to monitor and evaluate the quality of agricultural products in real time during the live-streaming sales process, thereby automating the quality inspection process and ensuring traceability reliability. Figure 1 This is a schematic diagram of the intelligent quality inspection system for live-streaming e-commerce according to an embodiment of the present invention, as shown below. Figure 1 As shown, the intelligent quality inspection system for live-streaming e-commerce according to an embodiment of the present invention specifically includes:

[0034] The multimodal data acquisition module 10 is used to acquire live streaming information during the live streaming process; the live streaming information specifically includes: live streaming image information, audio information, and / or text information;

[0035] The multimodal feature extraction and fusion module 11 is used to extract features from the live broadcast information and generate a fused feature representation of the live broadcast information through a multimodal fusion algorithm, thereby generating a multimodal fusion vector. Specifically, it is used to: extract image information features through a convolutional neural network (CNN), extract speech information features through an automatic speech recognition (ASR) model, and extract text information features through a natural language processing (NLP) technique; and fuse image information features, speech information features, and text information features using a Transformer or attention mechanism to generate a unified multimodal fusion vector.

[0036] The intelligent quality inspection and judgment module 12 is used to analyze the multimodal fusion vector through a trained deep learning model and output the quality inspection and evaluation results of the live-streamed product; the live-streamed product is an agricultural product.

[0037] The blockchain on-chain module 13 is used to upload the quality inspection and evaluation results and testing information to the blockchain to generate tamper-proof blockchain traceability information; the testing information specifically includes: testing timestamp, video clip summary, and / or hash value generated by packaging deep learning model version;

[0038] The visualization output module 14 is used to display the quality inspection and evaluation results and blockchain traceability information.

[0039] The system further includes:

[0040] The model training and update module is responsible for collecting data, training deep learning models, and updating and optimizing deep learning models based on new data and feedback.

[0041] The model storage and management module is used to store trained deep learning models and their version information, and to manage the access permissions and usage records of deep learning models.

[0042] The model invocation and execution module is used to invoke the corresponding deep learning model for inference and judgment when needed.

[0043] The technical solutions of the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The intelligent quality inspection system for live-stream e-commerce of agricultural products according to the embodiments of the present invention specifically includes:

[0044] (1) Multimodal data acquisition module: used to acquire image, voice and text information during the live broadcast of agricultural products;

[0045] (2) Multimodal feature extraction and fusion module: Based on convolutional neural network (CNN), speech recognition model (ASR) and natural language processing technology (NLP), image, speech and text features are extracted respectively, and fused feature representation is generated through multimodal fusion algorithm (such as Transformer or attention mechanism);

[0046] (3) Intelligent quality inspection and judgment module: Utilizes a trained deep learning model to analyze the fusion features and outputs evaluation results such as the quality grade and qualification status of agricultural products;

[0047] (4) Blockchain on-chain module: The quality inspection evaluation results, inspection timestamps, inspection model version hashes and other information are uploaded to the blockchain to form an immutable quality inspection record;

[0048] (5) Visual output module: Displays intelligent quality inspection reports and blockchain traceability information to consumers, platforms and regulators.

[0049] like Figure 2 As shown below, the functions of each module will be explained in detail.

[0050] Functional descriptions of each module:

[0051] 1. Multimodal data acquisition module, responsible for collecting image, voice and text information during the live broadcast of agricultural products.

[0052] 2. Multimodal feature extraction and fusion module: Based on CNN, ASR and NLP technologies, image, speech and text features are extracted respectively, and fused feature representation is generated through multimodal fusion algorithm.

[0053] 3. The intelligent quality inspection and judgment module uses a trained deep learning model to analyze the fused features and outputs evaluation results such as the quality grade and qualification status of agricultural products.

[0054] 4. The blockchain on-chain module uploads information such as quality inspection and evaluation results, inspection timestamps, and inspection model version hashes to the blockchain, forming an immutable quality inspection record.

[0055] 5. Visual output module, which displays intelligent quality inspection reports and blockchain traceability information to consumers, platforms and regulators.

[0056] 6. Deep learning model, the core model for intelligent quality inspection judgment, receives fused feature representations and performs analysis and judgment.

[0057] 7. The model training and update module is responsible for collecting data, training the model, and updating and optimizing the model based on new data and feedback.

[0058] 8. Model storage and management module: Stores trained models and their version information, and manages model access permissions and usage records.

[0059] 9. The model invocation and execution module invokes the corresponding deep learning model for inference and judgment when needed by the intelligent quality inspection and judgment module.

[0060] like Figure 3As shown, the modal data acquisition and fusion process is as follows:

[0061] 1. Image data acquisition:

[0062] Image data of agricultural products are collected in real time through cameras during live streaming.

[0063] 2. Image preprocessing:

[0064] Preprocessing operations such as denoising and enhancement are performed on the acquired images to improve image quality.

[0065] 3. Image feature extraction:

[0066] Features are extracted from preprocessed images using a convolutional neural network (CNN) model.

[0067] 4. Voice data acquisition:

[0068] The broadcaster's voice data is collected in real time via microphone.

[0069] 5. Speech preprocessing:

[0070] Preprocessing operations such as noise reduction and silence detection are performed on the collected speech.

[0071] 6. Speech Feature Extraction:

[0072] Features are extracted from preprocessed speech using an ASR (Automatic Speech Recognition) model.

[0073] 7. Text data collection:

[0074] Collect text data such as live stream captions and streamer descriptions.

[0075] 8. Text preprocessing:

[0076] Perform preprocessing operations on the text, such as word segmentation and stop word removal.

[0077] 9. Text Feature Extraction:

[0078] Features are extracted from preprocessed text using a Natural Language Processing (NLP) model.

[0079] 10. Multimodal Feature Fusion Module:

[0080] The Transformer or attention mechanism is used to fuse image, speech, and text features to generate a unified multimodal fusion vector.

[0081] 11. Fusion Feature Representation:

[0082] The fused feature representation is then passed to the intelligent quality inspection and judgment module.

[0083] 12. Intelligent quality inspection judgment:

[0084] Deep learning models are used to analyze and judge the fusion features, and output evaluation results such as the quality grade of agricultural products.

[0085] 13. Quality Inspection Result Output:

[0086] The quality inspection results are output in the form of a structured report for use by subsequent modules.

[0087] like Figure 4 As shown, the quality inspection and assessment process and the blockchain on-chain process specifically include:

[0088] 1. Multimodal fusion features:

[0089] Receive the fused feature vector from the multimodal feature fusion module.

[0090] 2. Intelligent quality inspection judgment:

[0091] Deep learning models are used to analyze and judge the fusion features, and output evaluation results such as the quality grade of agricultural products.

[0092] 3. Quality inspection results generation:

[0093] Based on the output of the intelligent quality inspection judgment module, a structured quality inspection result report is generated.

[0094] 4. Quality inspection result hash:

[0095] Hash algorithms such as SHA-256 are used to hash the quality inspection results to ensure the integrity and immutability of the data.

[0096] 5. Blockchain on-chain preparation:

[0097] The information, including the quality inspection result hash, timestamp, and model version hash, is packaged and prepared for writing to the blockchain.

[0098] 6. Blockchain on-chain:

[0099] The packaged data is written into the blockchain to form an immutable quality inspection record.

[0100] 7. Timestamp Records:

[0101] Record the current time when the quality inspection results are generated as part of the on-chain data on the blockchain.

[0102] 8. Model version hash:

[0103] Record the version number of the deep learning model used for quality inspection and perform hash processing to ensure the traceability of the model version.

[0104] 9. On-chain confirmation:

[0105] Wait for the blockchain network to confirm the written data to ensure that the data is successfully written to the block.

[0106] 10. Visual Presentation:

[0107] The quality inspection report and blockchain traceability information will be presented to consumers, platforms and regulators in a visual way.

[0108] 11. Blockchain traceability information:

[0109] It provides information such as block hashes on the blockchain for users to query and verify the authenticity of quality inspection records.

[0110] 12. On-chain completion:

[0111] The data has been successfully written to the blockchain, and the process is now complete.

[0112] Example 1: System Deployment

[0113] This system plugin is deployed on an e-commerce platform to automatically access live video streams, host audio, and product descriptions. In the background, the system extracts fruit images from the live stream in real time, analyzes the tapping sounds made by the host, and combines these with keywords used in the description (such as "no insect spots" and "plump") to generate multimodal feature vectors. After comprehensive evaluation by an AI model, it provides an assessment result stating, "The fruit has a complete appearance, solid sound quality, and credible language description; it is preliminarily judged as a superior product," and this result is written into the blockchain system.

[0114] Example 2: Blockchain Application

[0115] The system adopts a consortium blockchain architecture, with key participants including e-commerce platforms, livestreamers, consumers, and regulatory authorities. Each quality inspection process automatically generates a digest hash value, which, along with the version number of the detection model and the summary of the livestream segment, is uploaded to the blockchain, providing a reliable basis for subsequent consumer rights protection and regulatory audits.

[0116] Compared with the prior art, the technical solution of the embodiments of the present invention has the following beneficial effects:

[0117] High real-time performance: Supports real-time quality inspection and judgment during live streaming; Multimodal fusion: Integrates three information sources: image, voice, and text, improving recognition accuracy; Standardized output: Outputs structured quality inspection reports, facilitating understanding by regulators and consumers; Trustworthy evidence storage: Utilizes blockchain to ensure that quality inspection information is tamper-proof and traceable; Good scalability: The system supports rapid adaptation to different agricultural product types and platform access.

[0118] Method Implementation Examples

[0119] According to an embodiment of the present invention, a method for intelligent quality inspection of live-streaming e-commerce is provided, which is used in the aforementioned intelligent quality inspection system for live-streaming e-commerce. Figure 5 This is a flowchart of the intelligent quality inspection method for live-streaming e-commerce according to an embodiment of the present invention, such as... Figure 5 As shown, the intelligent quality inspection method for live-streaming e-commerce according to an embodiment of the present invention specifically includes:

[0120] Step S501: Collect live streaming information during the live streaming process through the multimodal data acquisition module; the live streaming information specifically includes: live streaming image information, audio information, and / or text information;

[0121] Step S502 involves extracting features from the live stream information using a multimodal feature extraction and fusion module, and generating a fused feature representation of the live stream information using a multimodal fusion algorithm to generate a multimodal fusion vector. Specifically, this includes: extracting image information features using a convolutional neural network (CNN), extracting speech information features using an automatic speech recognition (ASR) model, and extracting text information features using natural language processing (NLP) technology; and fusing the image information features, speech information features, and text information features using a Transformer or attention mechanism to generate a unified multimodal fusion vector.

[0122] Step S503: The intelligent quality inspection and judgment module uses a trained deep learning model to analyze the multimodal fusion vector and outputs the quality inspection and evaluation results of the live-streamed product; the live-streamed product is an agricultural product.

[0123] Step S504: The quality inspection and evaluation results and testing information are uploaded to the blockchain through the blockchain on-chain module to generate tamper-proof blockchain traceability information; the testing information specifically includes: testing timestamp, video clip summary, and / or hash value generated by packaging deep learning model version;

[0124] Step S505: Display the quality inspection and evaluation results and blockchain traceability information through the visualization output module.

[0125] The method further includes:

[0126] The model training and update module is responsible for collecting data, training deep learning models, and updating and optimizing deep learning models based on new data and feedback.

[0127] The model storage and management module stores trained deep learning models and their version information, and manages the access permissions and usage records of deep learning models.

[0128] The model invocation and execution module calls the corresponding deep learning model for inference and judgment when needed.

[0129] like Figure 3 As shown, the modal data acquisition and fusion process is as follows:

[0130] 1. Image data acquisition:

[0131] Image data of agricultural products are collected in real time through cameras during live streaming.

[0132] 2. Image preprocessing:

[0133] Preprocessing operations such as denoising and enhancement are performed on the acquired images to improve image quality.

[0134] 3. Image feature extraction:

[0135] Features are extracted from preprocessed images using a convolutional neural network (CNN) model.

[0136] 4. Voice data acquisition:

[0137] The broadcaster's voice data is collected in real time via microphone.

[0138] 5. Speech preprocessing:

[0139] Preprocessing operations such as noise reduction and silence detection are performed on the collected speech.

[0140] 6. Speech Feature Extraction:

[0141] Features are extracted from preprocessed speech using an ASR (Automatic Speech Recognition) model.

[0142] 7. Text data collection:

[0143] Collect text data such as live stream captions and streamer descriptions.

[0144] 8. Text preprocessing:

[0145] Perform preprocessing operations on the text, such as word segmentation and stop word removal.

[0146] 9. Text Feature Extraction:

[0147] Features are extracted from preprocessed text using a Natural Language Processing (NLP) model.

[0148] 10. Multimodal Feature Fusion Module:

[0149] The Transformer or attention mechanism is used to fuse image, speech, and text features to generate a unified multimodal fusion vector.

[0150] 11. Fusion Feature Representation:

[0151] The fused feature representation is then passed to the intelligent quality inspection and judgment module.

[0152] 12. Intelligent quality inspection judgment:

[0153] Deep learning models are used to analyze and judge the fusion features, and output evaluation results such as the quality grade of agricultural products.

[0154] 13. Quality Inspection Result Output:

[0155] The quality inspection results are output in the form of a structured report for use by subsequent modules.

[0156] like Figure 4 As shown, the quality inspection and assessment process and the blockchain on-chain process specifically include:

[0157] 1. Multimodal fusion features:

[0158] Receive the fused feature vector from the multimodal feature fusion module.

[0159] 2. Intelligent quality inspection judgment:

[0160] Deep learning models are used to analyze and judge the fusion features, and output evaluation results such as the quality grade of agricultural products.

[0161] 3. Quality inspection results generation:

[0162] Based on the output of the intelligent quality inspection judgment module, a structured quality inspection result report is generated.

[0163] 4. Quality inspection result hash:

[0164] Hash algorithms such as SHA-256 are used to hash the quality inspection results to ensure the integrity and immutability of the data.

[0165] 5. Blockchain on-chain preparation:

[0166] The information, including the quality inspection result hash, timestamp, and model version hash, is packaged and prepared for writing to the blockchain.

[0167] 6. Blockchain on-chain:

[0168] The packaged data is written into the blockchain to form an immutable quality inspection record.

[0169] 7. Timestamp Records:

[0170] Record the current time when the quality inspection results are generated as part of the on-chain data on the blockchain.

[0171] 8. Model version hash:

[0172] Record the version number of the deep learning model used for quality inspection and perform hash processing to ensure the traceability of the model version.

[0173] 9. On-chain confirmation:

[0174] Wait for the blockchain network to confirm the written data to ensure that the data is successfully written to the block.

[0175] 10. Visual Presentation:

[0176] The quality inspection report and blockchain traceability information will be presented to consumers, platforms and regulators in a visual way.

[0177] 11. Blockchain traceability information:

[0178] It provides information such as block hashes on the blockchain for users to query and verify the authenticity of quality inspection records.

[0179] 12. On-chain completion:

[0180] The data has been successfully written to the blockchain, and the process is now complete.

[0181] Device Example 1

[0182] This invention provides an electronic device, such as... Figure 6 As shown, it includes: a memory 60, a processor 62, and a computer program stored in the memory 60 and executable on the processor 62, wherein the computer program, when executed by the processor 62, performs the steps as described in the method embodiment.

[0183] Device Example 2

[0184] This invention provides a computer-readable storage medium storing an information transmission implementation program, which, when executed by a processor 62, performs the steps described in the method embodiment.

[0185] The computer-readable storage media described in this embodiment include, but are not limited to, ROM, RAM, disk, or optical disk.

[0186] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A live-streaming e-commerce intelligent quality inspection system, characterized in that, include: A multimodal data acquisition module is used to collect live streaming information during the live broadcast. The multimodal feature extraction and fusion module is used to extract features from the live broadcast information and generate a fused feature representation of the live broadcast information through a multimodal fusion algorithm, thereby generating a multimodal fusion vector. The intelligent quality inspection and judgment module is used to analyze the multimodal fusion vector through a trained deep learning model and output the quality inspection and evaluation results of the live broadcast product. The blockchain on-chain module is used to upload the quality inspection and evaluation results and testing information to the blockchain, generating tamper-proof blockchain traceability information on the blockchain. The visualization output module is used to display the quality inspection and evaluation results and blockchain traceability information.

2. The system according to claim 1, characterized in that, The live broadcast information specifically includes: live broadcast image information, audio information, and / or text information; The detection information specifically includes: detection timestamp, video clip summary, and / or hash value generated by packaging deep learning model versions; The products featured in the live stream are agricultural products.

3. The system according to claim 1, characterized in that, The multimodal feature extraction and fusion module is specifically used to: extract image information features through a convolutional neural network (CNN), extract speech information features through an automatic speech recognition (ASR) model, and extract text information features through natural language processing (NLP) technology. The Transformer or attention mechanism is used to fuse image information features, speech information features, and text information features to generate a unified multimodal fusion vector.

4. The system according to claim 1, characterized in that, The system further includes: The model training and update module is responsible for collecting data, training deep learning models, and updating and optimizing deep learning models based on new data and feedback. The model storage and management module is used to store trained deep learning models and their version information, and to manage the access permissions and usage records of deep learning models. The model invocation and execution module is used to invoke the corresponding deep learning model for inference and judgment when needed.

5. A smart quality inspection method for live-streaming e-commerce, characterized in that, The method for the intelligent quality inspection system for live-streaming e-commerce as described in any one of claims 1 to 4 specifically includes: Live streaming information is collected during the live broadcast using a multimodal data acquisition module; The features of the live broadcast information are extracted by the multimodal feature extraction and fusion module, and the fused feature representation of the live broadcast information is generated by the multimodal fusion algorithm to generate a multimodal fusion vector. The intelligent quality inspection and judgment module uses a trained deep learning model to analyze the multimodal fusion vector and outputs the quality inspection and evaluation results of the live streaming product. The quality inspection and evaluation results and testing information are uploaded to the blockchain through the blockchain on-chain module, generating tamper-proof blockchain traceability information on the blockchain. The quality inspection and assessment results and blockchain traceability information are displayed through a visualization output module.

6. The method according to claim 5, characterized in that, The live broadcast information specifically includes: live broadcast image information, audio information, and / or text information; The detection information specifically includes: detection timestamp, video clip summary, and / or hash value generated by packaging deep learning model versions; The products featured in the live stream are agricultural products.

7. The method according to claim 5, characterized in that, The features of the live broadcast information are extracted through the multimodal feature extraction and fusion module, and the fused feature representation of the live broadcast information is generated through the multimodal fusion algorithm. The generation of the multimodal fusion vector specifically includes: extracting image information features through the convolutional neural network (CNN), extracting speech information features through the speech recognition model (ASR), and extracting text information features through the natural language processing (NLP) technology; and using the Transformer or attention mechanism to fuse the image information features, speech information features, and text information features to generate a unified multimodal fusion vector.

8. The method according to claim 5, characterized in that, The method further includes: The model training and update module is responsible for collecting data, training deep learning models, and updating and optimizing deep learning models based on new data and feedback. The model storage and management module stores trained deep learning models and their version information, and manages the access permissions and usage records of deep learning models. The model invocation and execution module calls the corresponding deep learning model for inference and judgment when needed.

9. An electronic device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the steps of the intelligent quality inspection method for live-streaming e-commerce as described in any one of claims 5 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores an information transmission implementation program, which, when executed by a processor, implements the steps of the live-streaming e-commerce intelligent quality inspection method as described in any one of claims 5 to 8.

Images