Information evaluation method and device based on large model, intelligent agent, equipment, medium and product

Abstract

The invention provides an information evaluation method and device based on a large model, an intelligent agent, equipment, a medium and a product, and relates to the technical field of artificial intelligence, in particular to the fields of computer vision, deep learning, large models, health monitoring and the like. The information evaluation method based on the large model comprises the steps that in response to a received evaluation request for a first object, a first image indicated by the evaluation request is recognized, to-be-evaluated information is displayed on a first interface, the first image comprises a display interface image of target equipment, and the to-be-evaluated information is displayed on a second interface; the target device is used for collecting to-be-evaluated information of a target type of the first object, and the display interface image comprises the to-be-evaluated information; the to-be-evaluated information and the historical information of the first object with the target type are input to generate a large model, an information evaluation result is displayed on the first interface, and the information evaluation result is determined according to the relation between the to-be-evaluated information and the historical information.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technologies, and in particular to fields such as computer vision, deep learning, large models, and health monitoring. More specifically, it relates to an information evaluation method, apparatus, intelligent agent, device, medium, and product based on a large model. Background Art

[0002] The screen terminal can implement various interaction modes by integrating intelligent interaction, intelligent understanding, and content service technologies in the field of artificial intelligence (AI). In one example, the screen terminal can be used to manage user information and evaluate and display user information. Summary of the Invention

[0003] The present disclosure provides an information evaluation method, apparatus, intelligent agent, device, medium, and product based on a large model.

[0004] According to one aspect of the present disclosure, there is provided an information evaluation method based on a large model, including: in response to receiving an evaluation request for a first object, identifying a first image indicated by the evaluation request to display information to be evaluated on a first interface, where the first image is a display interface image of a target device, the target device is used to collect information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated; and inputting the information to be evaluated and historical information of the first object having the target type into an information generation large model to display an information evaluation result on the first interface, where the information evaluation result is determined according to the relationship between the information to be evaluated and the historical information.

[0005] According to another aspect of the present disclosure, there is provided an information evaluation apparatus based on a large model, including: an image recognition module, configured to, in response to receiving an evaluation request for a first object, identify a first image indicated by the evaluation request to display information to be evaluated on a first interface, where the first image is a display interface image of a target device, the target device is used to collect information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated; and an information evaluation module, configured to input the information to be evaluated and historical information of the first object having the target type into an information generation large model to display an information evaluation result on the first interface, where the information evaluation result is determined according to the relationship between the information to be evaluated and the historical information.

[0006] According to another aspect of the present disclosure, there is provided an intelligent agent of artificial intelligence, configured to execute the above-mentioned information evaluation method based on a large model.

[0007] According to another aspect of the present disclosure, there is provided an electronic device, including: one or more processors; a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0008] According to another aspect of the present disclosure, there is provided a computer-readable storage medium, on which computer programs or instructions are stored, and when the computer programs or instructions are executed by a processor, the steps of the above method are implemented.

[0009] According to another aspect of the present disclosure, there is provided a computer program product, including computer programs or instructions, and when the computer programs or instructions are executed by a processor, the steps of the above method are implemented.

[0010] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present disclosure, nor is it used to limit the scope of the present disclosure. Other features of the present disclosure will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] Through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, the above and other objects, features, and advantages of the present disclosure will become clearer. In the drawings:

[0012] Figure 1 Schematically shows a system architecture to which the information evaluation method based on a large model can be applied according to an embodiment of the present disclosure;

[0013] Figure 2 Schematically shows a flowchart of the information evaluation method based on a large model according to an embodiment of the present disclosure;

[0014] Figure 3A Schematically shows an example schematic diagram of a first interface during the process of adding a new candidate object according to an embodiment of the present disclosure;

[0015] Figure 3B Schematically shows an example schematic diagram of a third interface during the process of adding a new candidate object according to an embodiment of the present disclosure;

[0016] Figure 4 Schematically shows an example schematic diagram of a first interface after receiving an evaluation request for a first object according to an embodiment of the present disclosure;

[0017] Figure 5A Schematically shows an example schematic diagram of identifying a first image to display information to be evaluated on a first interface according to an embodiment of the present disclosure;

[0018] Figure 5BSchematically shown is an exemplary schematic diagram of recognizing a first image to display information to be evaluated on a first interface in accordance with another embodiment of the present disclosure;

[0019] Figure 6A Schematically shown is an exemplary schematic diagram of the process of displaying an information evaluation result on a first interface in accordance with an embodiment of the present disclosure;

[0020] Figure 6B Schematically shown is an exemplary schematic diagram of the process of displaying an information evaluation result on a first interface in accordance with an embodiment of the present disclosure;

[0021] Figure 6C Schematically shown is an exemplary schematic diagram of the process of displaying an information evaluation result on a first interface in accordance with another embodiment of the present disclosure;

[0022] Figure 7A Schematically shown is an exemplary schematic diagram of an interface corresponding to a push channel in accordance with an embodiment of the present disclosure;

[0023] Figure 7B Schematically shown is an exemplary schematic diagram of push information in accordance with an embodiment of the present disclosure;

[0024] Figure 7C Schematically shown is an exemplary schematic diagram of push information in accordance with another embodiment of the present disclosure;

[0025] Figure 8 Schematically shown is a block diagram of an information evaluation device based on a large model in accordance with an embodiment of the present disclosure;

[0026] Figure 9 Schematically shown is a structural block diagram of an agent of a large model in accordance with an embodiment of the present disclosure; and

[0027] Figure 10 Schematically shown is a block diagram of an electronic device suitable for implementing an information evaluation method based on a large model in accordance with an embodiment of the present disclosure. Detailed Description of the Invention

[0028] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. It should be understood, however, that these descriptions are merely exemplary and are not intended to limit the scope of the present disclosure. In the following detailed description, for the purpose of explanation, numerous specific details are set forth in order to provide a thorough understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may be practiced without these specific details. In addition, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts of the present disclosure.

[0029] The terms used herein are for describing specific embodiments only and are not intended to limit the present disclosure. The terms "comprising", "including" and the like as used herein indicate the presence of the recited features, steps, operations and / or components, but do not preclude the presence or addition of one or more other features, steps, operations or components.

[0030] All terms used herein (including technical and scientific terms) have the meanings commonly understood by those of ordinary skill in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted to have a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0031] In the case of using expressions such as "at least one of A, B, and C, etc.", generally, it should be interpreted according to the meaning commonly understood by those of ordinary skill in the art (for example, "a system having at least one of A, B, and C" should include, but is not limited to, a system having only A, only B, only C, having A and B, having A and C, having B and C, and / or having A, B, and C, etc.).

[0032] In one example, a screen terminal can be used for user health management. For example, the user can manage the measurement data of himself or others in a manual input manner on the screen terminal, so as to realize the attribution of the measurement data of the medical device. However, since this way of user health management based on the screen terminal requires the user to manually enter data, the degree of automation is relatively low.

[0033] Alternatively, a connection relationship between the screen terminal and the medical device can be established through Bluetooth, network configuration, etc. While the screen terminal and the medical device are in a connected state, the medical device is used for measurement and then the policy data is synchronized to the screen terminal. However, since this way of user health management based on the screen terminal requires pre-binding of the medical device and is usually limited to specific medical device brands and models, the medical device only feeds back the measurement data after measurement, and it is difficult to conduct detailed index analysis or personalized recommendation.

[0034] Therefore, an information evaluation scheme based on a large model is proposed in the embodiments of the present disclosure. For example, in response to receiving an evaluation request for a first object, the first image indicated by the evaluation request is recognized to display the information to be evaluated on a first interface, where the first image is a display interface image including a target device, the target device is used to collect the information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated; and the information to be evaluated and the historical information of the first object having the target type are input into an information generation large model to display an information evaluation result on the first interface, where the information evaluation result is determined according to the relationship between the information to be evaluated and the historical information.

[0035] According to an embodiment of the present disclosure, by recognizing a first image including the display interface of a target device, the information to be evaluated collected by the target device is obtained, so that the data intercommunication between the screen terminal and the target device no longer depends on prior binding and connection, which helps to improve the adaptability of the screen terminal. On this basis, by using a large model to analyze the information to be evaluated and the historical information of the first object having a target type, the information evaluation result can be obtained and displayed on the screen terminal, which helps to improve the accuracy and automation of information evaluation.

[0036] In the technical solution of the present invention, the processing of collection, storage, use, processing, transmission, provision, and disclosure of user personal information complies with the provisions of relevant laws and regulations and does not violate public order and good customs.

[0037] In the technical solution of the present invention, the authorization or consent of the user is obtained before obtaining or collecting the user personal information.

[0038] Figure 1 Schematically shows a system architecture to which the information evaluation method based on a large model according to an embodiment of the present disclosure can be applied. It should be noted that, Figure 1 What is shown is only an example of a system architecture to which the embodiments of the present disclosure can be applied, to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0039] As Figure 1 shown, the system architecture 100 according to this embodiment may include a first terminal device 101, a second terminal device 102, a third terminal device 103, a network 104, a first candidate device 105, a second candidate device 106, and a third candidate device 107. The network 104 is a medium for providing a communication link between different devices.

[0040] It should be noted that the information evaluation method based on a large model provided by the embodiments of the present disclosure can generally be executed by the first terminal device 101, the second terminal device 102, and the third terminal device 103. Correspondingly, the information evaluation device provided by the embodiments of the present disclosure can generally be arranged in the first terminal device 101, the second terminal device 102, and the third terminal device 103.

[0041] Alternatively, the information evaluation method based on a large model provided by the embodiments of the present disclosure can also be executed by the first candidate device 105, the second candidate device 106, and the third candidate device 107. Correspondingly, the information evaluation device provided by the embodiments of the present disclosure can also be arranged in the first candidate device 105, the second candidate device 106, and the third candidate device 107.

[0042] It should be understood that Figure 1 the numbers of the terminal devices, networks, and candidate devices in [[ ]] are merely illustrative. According to the implementation requirements, there can be any number of terminal devices, networks, and candidate devices.

[0043] It should be noted that the sequence numbers of the respective operations in the following methods are only used as representations of the operations for description purposes and should not be regarded as indicating the execution order of the respective operations. Unless explicitly stated, the method does not need to be executed exactly in the order shown.

[0044] Figure 2 The flowchart of the information evaluation method based on a large model according to an embodiment of the present disclosure is schematically shown.

[0045] As Figure 2 shown, the information evaluation method 200 based on a large model includes operations S210 to S220.

[0046] In operation S210, in response to receiving an evaluation request for a first object, by recognizing a first image indicated by the evaluation request, to display information to be evaluated on a first interface, where the first image is a display interface image including a target device, and the target device is used to collect information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated.

[0047] In operation S220, input the information to be evaluated and historical information of the first object having a target type into an information generation large model to display an information evaluation result on the first interface, where the information evaluation result is determined based on the relationship between the information to be evaluated and the historical information.

[0048] In an embodiment of the present disclosure, before obtaining the first image, consent or authorization of the first object may be obtained. For example, before operation S210, a request to obtain the first image may be sent to the first object. When the first object consents or authorizes the acquisition of user information, operation S210 is executed.

[0049] The target device refers to the device currently used by the first object, that is, the device used to collect the information to be evaluated. The display interface of the target device refers to the display screen of the target device itself, which is used to display the information to be evaluated of the first object having a target type collected. For example, the target device can be a medical device of any brand, any model, and any type. In this case, the display interface can be used to display medical information collected using the medical device. Specifically, if the target device is a sphygmomanometer, the information to be evaluated displayed on the display interface is the blood pressure value; if the target device is a glucometer, the information to be evaluated displayed on the display interface is the blood glucose value, etc.

[0050] An evaluation request refers to a request for the first object to analyze the information to be evaluated collected by the target device. The triggering method of the evaluation request can be configured according to actual business requirements and is not limited here. For example, after the first object uses the target device to collect the information to be evaluated, a dialog box can automatically pop up on the target device or the screen terminal, asking the first object whether to evaluate the information to be evaluated collected this time. If the first object clicks the "OK" control, an evaluation request is sent. Alternatively, the evaluation request can also be triggered in response to detecting a voice command issued by the first object.

[0051] In one example, the screen terminal and the target device can be pre-bound. In this case, since the connection between the screen terminal and the target device has been established, an access method can be used to transmit the information to be evaluated between the target device and the screen terminal. That is, after the target device collects the information to be evaluated, it can directly transmit the information to be evaluated to the screen terminal in an access way. For example, the access method can refer to Bluetooth.

[0052] In another example, the screen terminal and the target device may not be bound. In this case, since the connection between the screen terminal and the target device has not been established, a non-access method can be used to transmit the information to be evaluated between the target device and the screen terminal. That is, after the target device collects the information to be evaluated, the screen terminal can actively take a picture of the target device using the camera to obtain a first image. The non-access method can cover more scenarios than the access method, thus solving the problems existing in the access method.

[0053] The first image refers to an image containing the target device taken by the camera of the screen terminal. Further, when the screen terminal takes this first image, it is necessary to capture the display interface of the target device, that is, the first image includes the display interface image of the target device. For example, when the first object uses the target device, the target device can be placed within a preset distance from the screen terminal so that the screen terminal automatically takes a picture of the target device and its display interface to obtain the first image.

[0054] The preset distance can be determined according to at least one of the shooting ability of the camera and the target type of the target device. For example, if the shooting ability of the camera of the screen terminal is high, the preset distance can be set relatively far; if the shooting ability of the camera of the screen terminal is low, the preset distance can be set relatively close. Alternatively, if the font of the information to be evaluated corresponding to the target type is large, the preset distance can be set relatively far; if the font of the information to be evaluated corresponding to the target type is small, the preset distance can be set relatively close.

[0055] After the screen terminal side obtains the first image, it can recognize the first image to obtain the information to be evaluated displayed on the display interface of the target device, and display the information to be evaluated on the first interface of the screen terminal itself. The specific recognition method can be configured according to actual business requirements and is not limited here.

[0056] After the screen terminal obtains the information to be evaluated, it can obtain the historical information of the first object with the target type according to the target type to which the information to be evaluated belongs. For example, the database on the screen terminal side may store the candidate historical information of multiple candidate types of at least one candidate object. After receiving the evaluation request for the first object, it can match the candidate historical information in the database according to the object identifier of the first object and the target type of the information to be evaluated, so as to obtain the historical information of the first object with the target type.

[0057] After obtaining the historical information of the first object with the target type, the information to be evaluated and the historical information of the first object with the target type can be input into the information generation large model, so as to use the information generation large model to analyze the information to be evaluated and the historical information, and obtain and display the information evaluation result on the first interface on the screen terminal side. The form of the information evaluation result can be configured according to actual business requirements and is not limited here. For example, the information evaluation result can be in text form. In this case, the information evaluation result may include the health status and relevant suggestions of the first object. Alternatively, the information evaluation result can be in chart form. In this case, the information evaluation result may include the numerical relationship between the information to be evaluated and the historical information, the change trend of the data, etc.

[0058] It should be noted that the display methods of the above information to be evaluated and the information evaluation result on the first interface of the screen terminal can be configured according to actual business requirements and are not limited here. For example, the screen terminal can display the information to be evaluated and the information evaluation result on the first interface in the form of a pop-up window. Alternatively, a fixed display area can be preset on the first interface of the screen terminal, and the information to be evaluated and the information evaluation result can be displayed in this display area.

[0059] According to the embodiments of the present disclosure, by recognizing the first image including the display interface of the target device, the information to be evaluated collected by the target device is obtained, so that the data intercommunication between the screen terminal and the target device no longer depends on the need to be bound and connected in advance, which helps to improve the adaptability of the screen terminal. On this basis, by using the large model to analyze the information to be evaluated and the historical information of the first object with the target type, the information evaluation result can be obtained and displayed on the screen terminal, which helps to improve the accuracy and automation degree of information evaluation.

[0060] Figure 3ASchematically shows an example schematic diagram of a first interface during the process of adding a new candidate object according to an embodiment of the present disclosure.

[0061] In one example, the trigger information may refer to information used to initiate an operation of adding an object. For example, the trigger information may be obtained in response to a user clicking a button on a smart terminal; alternatively, the trigger information may also be obtained when the smart terminal detects that a certain specific condition is met; alternatively, the trigger information may also be obtained in response to a user's voice command.

[0062] As Figure 3A shown, in 300A, the first interface of the smart terminal can be used to configure the device user. Currently, the configured device users include candidate object 1 and candidate object 2. The user can click the object addition button on this first interface to obtain the trigger information to add a device user.

[0063] Figure 3B Schematically shows an example schematic diagram of a third interface during the process of adding a new candidate object according to an embodiment of the present disclosure.

[0064] As Figure 3B shown, in 300B, after the smart terminal receives the trigger information, it can display the third interface and use the third interface to obtain attribute information for the candidate object. The attribute information refers to the characteristic information of the candidate object. For example, age, gender, contact information, etc. For example, the user can click the image and gender button on this third interface to input the image and gender of the newly added candidate object 3. Alternatively, the user can input the attribute information by clicking the corresponding attribute item on this third interface.

[0065] In addition, the user can also use the third interface to obtain auxiliary information for the candidate object. The auxiliary information refers to information used to further identify and verify the identity of the candidate object. For example, the auxiliary information may include at least one of the following: visual information and voiceprint information. Visual information refers to data related to vision, such as photos, videos, etc. of the candidate object. Voiceprint information refers to information related to sound, such as a piece of sound recorded by the candidate object reading specific text.

[0066] The input methods of the above-mentioned attribute information and auxiliary information can be configured according to actual business requirements and are not limited herein. For example, the attribute information and auxiliary information can be manually input by the user through controls such as input boxes and selection boxes provided on the third interface. Alternatively, the attribute information and auxiliary information can also be obtained from files imported through the third interface.

[0067] After receiving the trigger information, a candidate identifier for uniquely identifying the candidate object can also be generated. For example, a user name, an ID number, etc. On this basis, the candidate identifier, visual information, and auxiliary information of the candidate object can be associated and stored in the object mapping. The object mapping refers to a storage structure for associating the candidate identifier with the attribute information and auxiliary information of the candidate object to facilitate subsequent querying and management. For example, the object mapping can be a database table, a key-value store, a file system, etc.

[0068] According to an embodiment of the present disclosure, by responding to the trigger information, obtaining the attribute information and auxiliary information of the candidate object through the third interface, and associating and storing these information with the candidate identifier in the object mapping, the comprehensive collection and efficient management of the information of the candidate object are achieved, ensuring the accuracy and integrity of the information.

[0069] In one example, after receiving the evaluation request, the auxiliary information for the first object can be determined according to the object identifier of the first object and the object mapping. The object mapping can include the corresponding relationships between multiple groups of candidate identifiers, attribute information, and auxiliary information. For example, the object identifier of the first object can be matched with each candidate identifier in the object mapping, and the auxiliary information corresponding to the candidate identifier that matches the object identifier can be determined as the auxiliary information for the first object.

[0070] The auxiliary information for the first object can include at least one of the following: visual information and voiceprint information. The evaluation request can include input information. The input information can include at least one of the second image of the first object and the audio to be evaluated of the first object. After determining the auxiliary information for the first object, the input information can be matched with the auxiliary information to determine whether they match.

[0071] The specific matching method and the discrimination condition for whether they match can be configured according to the actual business requirements and are not limited herein. For example, the matching can be to match the second image with the visual information, and / or the matching can also be to match the audio to be evaluated with the voiceprint information. The discrimination condition can be that the first image matches the visual information, and / or the audio to be evaluated matches the voiceprint information.

[0072] After determining that the input information matches the auxiliary information, the information to be evaluated and the information evaluation result can be updated to the object mapping as the historical information of the first object, so that the information to be evaluated and the information evaluation result are attributed to the corresponding first object. It should be noted that the above information to be evaluated and the information evaluation result can be attributed through the aforementioned automatic recognition method, or can also be attributed through the interaction between the user and the interface.

[0073] According to an embodiment of the present disclosure, by managing the correspondence between multiple groups of candidate identifiers, attribute information, and auxiliary information through object mapping, and using the matching between input information and auxiliary information in an evaluation request, accurate identification of the first object and update of historical information are achieved, which helps to improve the efficiency and accuracy of subsequent information evaluation for the first object.

[0074] Figure 4 Schematically shows an example schematic diagram of a first interface after receiving an evaluation request for a first object according to an embodiment of the present disclosure.

[0075] As Figure 4 shown, in 400, after determining that the input information matches the auxiliary information, a prompt message can be displayed on the first interface. For example, the prompt message can be "The target device is in use. Please ensure that the target device is worn and started. Please wait patiently for the synchronization of data after the measurement of the target device is completed."

[0076] It should be noted that during the process of the first object using the target device, the target device needs to be within a preset distance from the smart screen to ensure that the camera of the smart screen can capture the first image of the target device.

[0077] Figure 5A Schematically shows an example schematic diagram of identifying a first image according to an embodiment of the present disclosure to display information to be evaluated on a first interface.

[0078] In one example, the morphological features of multiple candidate devices can be obtained in advance. The morphological features refer to visual recognizable features such as the appearance shape, size, color, button layout, display position and shape of the candidate device. For example, the morphological feature of candidate device A is that the display screen is rectangular and located in the center of the front of candidate device A, and there are three circular operation buttons beside it; the morphological feature of candidate device B is that the display screen is square and located at the top of the device, and there is a long strip-shaped notch beside it, etc.

[0079] As Figure 5A shown, in 500A, after receiving the first image 501, the target device 502 in the first image 501 can be classified according to the morphological features 503 of multiple candidate devices to obtain a target type 504. The specific implementation manner of the classification can be configured according to actual business requirements and is not limited herein.

[0080] For example, the morphological features 503 of multiple candidate devices can be used to train a convolutional neural network, that is, using the morphological feature images of the candidate devices as inputs and the device type labels as outputs. In this case, the trained convolutional neural network is used to process the first image 501. The convolutional neural network automatically extracts features from the first image 501 and compares these features with the morphological features 503 of multiple candidate devices respectively, so as to classify the target device 502.

[0081] Alternatively, the morphological features 503 of multiple candidate devices can be used to train a support vector machine, that is, using the morphological feature images of the candidate devices as inputs and the device type labels as outputs. In this case, the trained support vector machine is used to process the first image 501. The support vector machine automatically extracts features from the first image 501 and compares these features with the morphological features 503 of multiple candidate devices respectively, so as to classify the target device.

[0082] After determining the target type 504, text recognition can be performed on the first image 501 according to the target type 504 to obtain and display the information to be evaluated 506 on the first interface. The specific method of text recognition can be configured according to actual business requirements and is not limited here. For example, dedicated optical character recognition models can be trained respectively for the display screen characteristics of different candidate types of devices, such as font, size, arrangement method, etc.; then after determining the target type 504, the corresponding optical character recognition model is used for text recognition.

[0083] Alternatively, according to the target type 504, the text area and format to be recognized can be determined. On this basis, by intercepting the image of the text area in the display interface image, a regional image 505 is obtained. The regional image 505 refers to the partial image of the first image 501 that displays the information to be evaluated 506; then the optical character recognition technology is used to perform text recognition on the regional image 505 to extract the text content to obtain the information to be evaluated 506.

[0084] It should be noted that the specific method of intercepting the regional image can be configured according to actual business requirements and is not limited here. For example, the intercepting method can include at least one of the following: the intercepting method based on template matching, the object detection intercepting method based on deep learning, and the intercepting method based on image segmentation, etc.

[0085] The intercepting method based on template matching means that the display screen templates of devices of different target types are stored in advance. The display screen templates can include, for example, features such as the shape, size, and position of the templates. When the target type is determined, according to the display screen template of the device of this target type, template matching is performed in the first image 501 to find the region most similar to the display screen template, so as to intercept and obtain the regional image 505.

[0086] The object detection-based cropping method refers to using the object detection algorithm in deep learning to train a large number of images of target type devices with the positions of the display screens marked, and directly detecting the position of the display screen of the target device in the first image 501 with the trained model, and cropping it out as the region image 505.

[0087] The cropping method based on image segmentation refers to using image segmentation technology to divide different regions in the first image 501, and then determining the segmented region where the display screen is located according to the characteristics of the display screen of the target type device, such as color, texture, etc., and further cropping to obtain the region image 505.

[0088] According to the embodiments of the present disclosure, by first accurately cropping the display screen area of the target device according to the target type and then performing text recognition, different cropping and recognition methods can be flexibly adopted according to the characteristics of the display screens of different types of medical devices, thereby effectively improving the accuracy and efficiency of the recognition of the information to be evaluated.

[0089] According to the embodiments of the present disclosure, by first classifying the target device according to the morphological characteristics of the candidate device to determine the target type, and then performing text recognition and display on the first image according to the target type, accurate and rapid recognition and presentation of the measurement data of different devices are realized, so as to be able to effectively adapt to a variety of devices.

[0090] Figure 5B Schematically shows an example schematic diagram of the process of recognizing the first image to display the information to be evaluated on the first interface according to another embodiment of the present disclosure.

[0091] In one example, a recognition model can be pre-trained using training samples. The training samples can cover various candidate devices, images affected by multiple perspective parameters and multiple environmental parameters. The candidate devices can refer to devices of various brands, various models and various forms. For example, taking the candidate device as a medical device, the candidate devices can include sphygmomanometers, blood glucose meters, pulse oximeters, etc. of different brands and different models.

[0092] The perspective parameter refers to the parameter of the relative angle and relative distance between the shooting device and the candidate device. For example, the perspective parameter can include shooting at 30 cm from the front of the candidate device, or at a 45-degree angle from the side of the candidate device, etc. The environmental parameter refers to the environmental factor parameter that affects the image shooting effect, such as light intensity, light color, background complexity, etc. For example, shooting under bright indoor light, or shooting under dim light conditions, or shooting in an environment with complex background patterns, etc.

[0093] It should be noted that if the amount of data in the training samples is small, data augmentation techniques can also be used to expand the training samples. That is, data augmentation operations can be performed on the original training samples, such as rotation, scaling, brightness adjustment, adding noise, etc., to simulate images under different perspective parameters and environmental parameters, expand the quantity and diversity of the training samples, and improve the generalization ability of the model.

[0094] In one example, the training samples can include a plurality of positive sample pairs and a plurality of negative sample pairs. The positive sample pairs and the negative sample pairs are dynamically updated at predetermined time intervals, and the number of positive sample pairs and the number of negative sample pairs satisfy a predetermined balance condition.

[0095] The sample images included in the positive sample pairs match the label types of the candidate devices. For example, a positive sample pair can be a sample image 1 including device A, and the label type is device A. The sample images included in the negative sample pairs do not match the label types. For example, a negative sample pair can be a sample image 1 including device A, and the label type is device B.

[0096] In one example, update operations can be performed on the positive sample pairs and negative sample pairs in the training samples at preset time intervals, such as adding new sample pairs, deleting old sample pairs, etc., to maintain the timeliness and representativeness of the training samples. For example, the training samples are updated once a week, new sample pairs are added, or some outdated or poor-quality sample pairs are deleted.

[0097] In another example, the predetermined balance condition refers to a condition preset to ensure the balance between the number of positive sample pairs and negative sample pairs, so as to prevent biases caused by uneven sample numbers during the model training process. After determining the number of positive sample pairs and the number of candidate negative sample pairs, the positive sample pairs and the candidate negative sample pairs can be screened according to the predetermined balance condition, so that the number of positive sample pairs and negative sample pairs satisfies the predetermined balance condition. The predetermined balance condition can be configured according to actual business requirements and is not limited herein. For example, the predetermined balance condition can be configured such that the number of positive sample pairs is equal to the number of negative sample pairs. Alternatively, the predetermined balance condition can be configured such that the difference between the number of positive sample pairs and the number of negative sample pairs is less than or equal to a predetermined threshold. The predetermined threshold can be set to 1.

[0098] According to the embodiments of the present disclosure, since the sample images included in the positive sample pairs match the label types of the candidate devices, and the sample images included in the negative sample pairs do not match the label types, by training the model using a plurality of positive sample pairs and a plurality of negative sample pairs, joint training of positive and negative samples is achieved, so that the obtained model can automatically perform device type classification, thereby improving the accuracy of target type determination.

[0099] The recognition model may include a device recognition network and a text recognition network. The device recognition network can be used to classify the devices in the first image, that is, classify the target device in the first image according to the morphological features of each of the multiple candidate devices to obtain the target type. The text recognition network can be used to recognize the text on the screen of the device in the first image, that is, perform text recognition on the first image according to the target type to obtain the information to be evaluated.

[0100] The specific structures of the device recognition network and the text recognition network can be configured according to actual business requirements and are not limited herein. For example, a convolutional neural network can be selected for the device recognition network, and a recurrent neural network can be selected for the text recognition network. Alternatively, a support vector machine, a random forest, etc. can also be selected for the device recognition network, and a convolutional neural network can also be selected for the text recognition network.

[0101] As Figure 5B shown, in 500B, after obtaining the first image 507, the captured first image 507 can be directly input into the pre-trained recognition model 509. The model 509 simultaneously performs device recognition and text recognition. After obtaining the information 511 to be evaluated, it is displayed through the first interface of the smart screen. Alternatively, the device recognition network 509_1 can be first used to recognize the first image 507 to obtain the target type 510, and then the text recognition network 509_2 can be used to perform text recognition on the image to obtain the information 511 to be evaluated, and finally it is displayed on the first interface.

[0102] According to the embodiments of the present disclosure, since the recognition model is trained with training samples including different perspectives and environmental parameters, it has good adaptability and robustness. On this basis, by using the pre-trained recognition model, through the cooperation of the device recognition network and the text recognition network, the medical device in the first image can be accurately recognized and text extracted, the information to be evaluated is obtained and displayed, making the entire recognition process efficient and accurate.

[0103] Figure 6A Schematically shows an example schematic diagram of the process of displaying the information evaluation result on the first interface according to the embodiments of the present disclosure.

[0104] In one example, the information evaluation result may include the evaluation level to which the information to be evaluated belongs. The determination method of the evaluation level can be configured according to actual business requirements and is not limited herein. It should be noted that the determination methods of the evaluation levels for different target types are different. For example, if the target device is a medical device and both the information to be evaluated and the historical information are index measurement values, the evaluation levels may include normal, mildly abnormal, moderately abnormal, severely abnormal, etc.

[0105] Exemplarily, the evaluation level can be determined based on the target type according to the numerical relationship between the information to be evaluated and the historical information, and is used to characterize the position of the information to be evaluated in the historical information. For example, for each target type, the index range can be pre-divided for each evaluation level according to the historical information of this target type. It should be noted that the division of the index range is associated with the attribute information of the first object, that is, for the same target type, objects with different attribute information may correspond to different index ranges. The attribute information can include, for example, age, gender, etc.

[0106] After obtaining the information to be evaluated, the information to be evaluated can be compared with the index range determined based on the historical information to obtain a deviation value. The deviation value characterizes the deviation degree of the information to be evaluated relative to the index range and is used to quantify the deviation of the information to be evaluated from the normal range. After obtaining the deviation value, the evaluation level is determined according to the target type and the deviation value.

[0107] The calculation method of the deviation value can be configured according to the actual business requirements and is not limited here. For example, the deviation value can be obtained by directly comparing the information to be evaluated with the upper and lower limits of the index range; alternatively, the deviation value can also be obtained by calculating the percentage deviation of the information to be evaluated relative to the central value of the index range; alternatively, the deviation value can also be obtained by calculating the mean and standard deviation of the index range based on the historical information, and then calculating how many standard deviations the difference between the information to be evaluated and the mean is.

[0108] Alternatively, the evaluation level can be determined by analyzing the distribution of the historical information and according to the position of the information to be evaluated in the historical information. For example, the percentile of the information to be evaluated in the history can be calculated. If it is in the top 20%, it is normal; in the middle 50%, it is mildly abnormal; in the bottom 30%, it is severely abnormal, etc.

[0109] According to the embodiments of the present disclosure, by comparing the information to be evaluated with the index range determined based on the attribute information to obtain a deviation value, and determining the evaluation level according to the target type and the deviation value, considering the attribute differences of different objects, the evaluation result is more in line with the individual actual situation, realizing personalized information evaluation. The accuracy and pertinence of the evaluation are improved.

[0110] After obtaining the evaluation level, the evaluation level can be displayed on the first interface according to the preset display method for the target type. It should be noted that the preset display methods for the evaluation levels of different target types are different. For example, the preset display method can include at least one of the following: color, icon, text format, etc.

[0111] Exemplarily, for different target types, different colors can be preset for different evaluation levels, and the corresponding colors are used to display the evaluation levels on the first interface. For example, for the blood glucose meter target type, when the evaluation level is "normal", it is displayed in green, "mild abnormality" in yellow, and "severe abnormality" in red.

[0112] Alternatively, for different target types, different icon identifiers can be preset for different evaluation levels, and the corresponding icon identifiers are used to display the evaluation levels on the first interface. For example, when the evaluation level is "normal", a smiling face icon is displayed, a calm face icon for "mild abnormality", and a crying face icon for "severe abnormality".

[0113] Alternatively, for different target types, different text formats can be preset for different evaluation levels, and the corresponding text formats are used to display the evaluation levels on the first interface. For example, when the evaluation level is "normal", it is displayed in a regular font, in bold for the "warning" level, and in a larger bold font with flashing for the "danger" level, etc.

[0114] As Figure 6A shown, in 600A, the first interface displays the information 1 to be evaluated of type 1 of the first object, and the corresponding evaluation level is displayed by color; it also displays the information 2 to be evaluated of type 2 of the first object, and the corresponding evaluation level is displayed by color; it displays the information 3 to be evaluated of type 3 of the first object, and the corresponding evaluation level is displayed by color; it displays the information 4 to be evaluated of type 4 of the first object, and the corresponding evaluation level is displayed by color.

[0115] For the above types 1 to 4, since the evaluation level corresponding to type 1 is "severe abnormality", it can also be highlighted by an exclamation mark icon identifier. The user can access the details of type 1 by clicking on the corresponding part in the first interface.

[0116] According to the embodiments of the present disclosure, by combining the evaluation level in the information evaluation result with the target type and adopting an evaluation level determination method based on the relationship between the target type and the information to be evaluated and the historical information, the accuracy and personalization degree of the evaluation level are improved. On this basis, by intuitively displaying the evaluation level on the first interface according to the preset display method, it can effectively help the user quickly understand the trend of the information to be evaluated compared with the historical information.

[0117] In one example, after obtaining the evaluation level, it is also possible to determine whether the evaluation level conforms to a predetermined response rule. If the evaluation level conforms to the predetermined response rule, warning information can be pushed to a second object based on the push channel corresponding to the target type. The warning information can be used to inform the evaluation situation of the information to be evaluated of the first object. For example, the warning information can be "The blood glucose value of the user you are concerned about is abnormal. Please contact in time!".

[0118] The predetermined response rule refers to a rule preset for determining what response measures should be taken at different evaluation levels. The predetermined response rule can define candidate channels and candidate objects for different evaluation levels of each candidate type.

[0119] The candidate channel refers to, for each candidate type, a set of various channels preset for pushing warning information. The candidate channels can include at least one of text messages, applications, phone calls, emails, etc. For example, when the candidate type is blood glucose, the candidate channels can include text messages and in-application notifications; when the candidate type is blood pressure, the candidate channels include phone calls and emails.

[0120] The candidate object refers to, for each candidate type and different evaluation levels, a set of objects preset for receiving warning information. The candidate objects can include at least one of relatives, digital human doctors, and emergency contacts. For example, when the candidate type is blood glucose and the evaluation level is "moderately abnormal", the candidate object can be the relative of the user; when the candidate type is blood pressure and the evaluation level is "severely abnormal", the candidate object can be the user's digital human doctor or emergency contact.

[0121] According to an embodiment of the present disclosure, by pushing warning information to a second object based on the push channel corresponding to the target type when the evaluation level conforms to the predetermined response rule, the definition of the predetermined response rule improves the organization and pertinence of the warning, and can convey the warning information to the appropriate object according to different target types and different evaluation levels, realizing the timely response and effective transmission of the evaluation risk, and improving the efficiency and effect of information management.

[0122] In one example, the information evaluation result can include an evaluation description for the information to be evaluated. Taking the target device as a medical device as an example, the evaluation description can include index analysis and personalized diet and exercise suggestions, etc.

[0123] Figure 6B A schematic diagram shows an example of the process of displaying the information evaluation result on the first interface according to an embodiment of the present disclosure.

[0124] As Figure 6BAs shown, in 600B, the first interface can display details of type 1. For example, the details of type 1 can include the information to be evaluated 1 corresponding to type 1, the trend of type 1 obtained through index analysis, the analysis description, etc.

[0125] After obtaining the information to be evaluated, a large information generation model can be used to process the first prompt information for the target type and the information to be evaluated to obtain an evaluation description. The first prompt information can be used to guide the large information generation model to analyze the information to be evaluated to obtain an evaluation description.

[0126] The generation method of the first prompt information can be configured according to actual business needs and is not limited here. In one example, the first prompt information can be fixed, that is, a fixed first prompt information is preset for each target type and directly called during analysis. For example, for the target type of blood glucose, the first prompt information can be "Please analyze the user's blood glucose control situation based on the blood glucose value and give detailed dietary suggestions". In another example, the first prompt information can be dynamically generated, that is, the first prompt information is dynamically generated according to the real-time situation of the target type and the information to be evaluated. For example, when the information to be evaluated is blood glucose values for consecutive days, the dynamically generated first prompt information can be "Please analyze the change trend of the blood glucose values for multiple days and give a detailed evaluation description".

[0127] Figure 6C A schematic diagram showing an example of the process of displaying the information evaluation result on the first interface according to another embodiment of the present disclosure is schematically shown.

[0128] As Figure 6C shown, in 600C, the first interface can also display details of the first object. The first interface can provide a time period selection item and a type selection item. The user can view the details of the first object for different types and different time periods on the first interface by clicking on different time period selection items and different type selection items. The details can include the trend of type 1 obtained through index analysis and the recommended description obtained through the large information generation model, etc.

[0129] For example, taking the information to be evaluated as blood pressure value as an example, the recommended description can be "Your blood pressure value is within the normal range, but it has fluctuated greatly recently. It is recommended to pay attention to rest, reduce stress, and maintain good living habits".

[0130] In one example, the predefined response rule can also define candidate templates for each candidate type. A candidate template refers to a format designed in advance for different types of evaluation information to generate corresponding prompt information. For example, the candidate template can contain some variable placeholders to be filled according to specific situations to generate personalized information.

[0131] In cases where an evaluation description needs to be generated, second prompt information can also be generated based on candidate templates for the target type. The second prompt information is used to guide the information generation large model to further interact with the user to obtain more supplementary information, so as to more accurately complete the evaluation description. For example, the second prompt information may be a question such as "Your blood glucose value shows some fluctuations. Have you adjusted your diet or exercise habits recently?"

[0132] After obtaining the second prompt information, based on the second prompt information, the information generation large model can conduct multiple rounds of interactive conversations with the first object to enable the first object to provide supplementary information. Multiple rounds of interactive conversations refer to continuous multi-round dialogue interactions between the information generation large model and the user. During this process, new prompt information can be generated as needed to guide the user to provide more detailed or accurate information for better task completion. Supplementary information refers to the sum of all information generated during multiple rounds of interactive conversations. For example, the user answered information such as increasing sweet food intake and reducing exercise volume recently during multiple rounds of interactive conversations.

[0133] After obtaining the supplementary information, the information generation large model can be used to process the supplementary information, the first prompt information, and the information to be evaluated obtained through multiple rounds of interactive conversations to obtain an evaluation description.

[0134] According to an embodiment of the present disclosure, by combining predetermined response rules, candidate templates, and multiple rounds of interactive conversations, the interactivity between the user and the system is enhanced. Appropriate templates can be automatically selected according to different types of health data, guiding the user to supplement necessary information, and using the information generation model to generate accurate and detailed evaluation descriptions, realizing in-depth evaluation of the information to be evaluated and generation of personalized suggestions.

[0135] According to an embodiment of the present disclosure, by using the information generation large model to generate an evaluation description based on the first prompt information and the information to be evaluated, due to the combination of the target type and the prompt information, the pertinence and accuracy of the evaluation description are ensured, the generation efficiency of the evaluation description is improved, and a detailed and personalized explanation of the information to be evaluated is realized.

[0136] In one example, each candidate object in the object mapping can be managed based on candidate accounts. For example, one candidate account can correspond to at least one candidate object. For each first object, the candidate accounts in the object mapping can be matched according to the target account to which the first object belongs to determine at least one candidate object corresponding to the target account. If there are other candidate objects other than the first object among the at least one candidate object, it means that the first object has associated objects; if there are no other candidate objects other than the first object among the at least one candidate object, it means that the first object has no associated objects.

[0137] When it is determined that the first object has an associated object, a push message can be generated at intervals of a predetermined time period based on the historical information corresponding to the first object in the object mapping. The predetermined time period can be, for example, one week, one month, three months, etc. Taking each candidate device as a medical device as an example, the push message can be a detailed health report generated by summarizing and analyzing various index records. After obtaining the push message, the push message can be pushed to the associated object based on the push channel corresponding to the associated object. The push channel can include at least one of SMS, email, App notification, etc.

[0138] Figure 7A Schematically shows an example schematic diagram of an interface corresponding to a push channel according to an embodiment of the present disclosure.

[0139] As Figure 7A shown, in 700A, the associated object can receive the push message about the first object through the interface corresponding to the push channel. In this case, the associated object can access the specific information details by clicking on the push message about the first object in the interface corresponding to the push channel.

[0140] Figure 7B Schematically shows an example schematic diagram of a push message according to an embodiment of the present disclosure.

[0141] As Figure 7B shown, in 700B, the interface corresponding to the push channel can provide type selection items. The user can view the details of the first object for different types by clicking on different type selection items in the interface corresponding to the push channel. The details can include the information to be evaluated 1 corresponding to type 1, the trend of type 1 obtained through index analysis, and the recommended description obtained through the information generation large model, etc.

[0142] Figure 7C Schematically shows an example schematic diagram of a push message according to another embodiment of the present disclosure.

[0143] As Figure 7C shown, in 700C, the interface corresponding to the push channel can also provide time period selection items and type selection items. The user can view the details of the first object for different types and different time periods by clicking on different time period selection items and different type selection items in the first interface. The details can include the trend of type 1 obtained through index analysis and the recommended description obtained through the information generation large model, etc.

[0144] For example, taking the information to be evaluated as the blood pressure value as an example, the recommended description can be "Your blood pressure value is within the normal range, but there are significant fluctuations recently. It is recommended to pay attention to rest, reduce stress, and maintain good living habits."

[0145] According to an embodiment of the present disclosure, by generating push information based on historical information in a predetermined period and sending the push information through a push channel corresponding to the associated object, it helps to ensure that the associated object can understand the situation of the first object in a timely manner, thereby improving the reliability and flexibility of information delivery.

[0146] The above are merely exemplary embodiments, but are not limited thereto. Other large model-based information evaluation methods known in the art may also be included, as long as they can improve the accuracy and automation of information evaluation.

[0147] Figure 8 The block diagram of the large model-based information evaluation device according to an embodiment of the present disclosure is schematically shown.

[0148] like Figure 8 As shown, the large model-based information evaluation device 800 may include an image recognition module 810 and an information evaluation module 820 .

[0149] The image recognition module 810 is used to respond to receiving an evaluation request for a first object by identifying the first image indicated by the evaluation request to display the information to be evaluated on a first interface, wherein the first image is a display interface image including a target device, the target device is used to collect information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated.

[0150] The information evaluation module 820 is used to generate a large model by inputting the information to be evaluated and the historical information of the first object with the target type to display the information evaluation result on the first interface, wherein the information evaluation result is determined based on the relationship between the information to be evaluated and the historical information.

[0151] According to an embodiment of the present disclosure, the image recognition module 810 may include a classification unit and a first recognition unit.

[0152] The classification unit is configured to classify the target device in the first image according to the morphological features of each of the plurality of candidate devices to obtain a target type.

[0153] The first recognition unit is configured to perform text recognition on the first image according to the target type, and obtain and display information to be evaluated on the first interface.

[0154] According to an embodiment of the present disclosure, the first identification unit may include an interception subunit and an identification subunit.

[0155] The interception subunit is used to intercept the information to be evaluated in the display interface image according to the target type to obtain a regional image, wherein the regional image is a portion of the first image that displays the information to be evaluated.

[0156] An identification subunit for performing text identification on a regional image to obtain information to be evaluated.

[0157] According to an embodiment of the present disclosure, the image recognition module 810 may include a second recognition unit.

[0158] The second recognition unit is configured to use a pre-trained recognition model to recognize the first image, obtain and display the information to be evaluated on the first interface, where the recognition model is obtained by training with training samples, and the training samples include images of multiple candidate devices under the influence of different perspective parameters and different environmental parameters.

[0159] According to an embodiment of the present disclosure, the recognition model includes a device recognition network and a text recognition network. The device recognition network is configured to classify the target device in the first image according to the morphological features of multiple candidate devices to obtain a target type, and the text recognition network is configured to perform text recognition on the first image according to the target type to obtain the information to be evaluated.

[0160] According to an embodiment of the present disclosure, the training samples include multiple positive sample pairs and multiple negative sample pairs. The positive sample pairs and the negative sample pairs are dynamically updated at intervals of a predetermined time period, and the number of positive sample pairs and the number of negative sample pairs satisfy a predetermined balance condition. The sample images included in the positive sample pairs match the label types of the candidate devices, and the sample images included in the negative sample pairs do not match the label types.

[0161] According to an embodiment of the present disclosure, the information evaluation result includes the evaluation level to which the information to be evaluated belongs.

[0162] According to an embodiment of the present disclosure, the information evaluation device 800 based on a large model may further include a first determination module and a display module.

[0163] The first determination module is configured to determine the evaluation level based on the relationship between the information to be evaluated and the historical information according to the target type.

[0164] The display module is configured to display the evaluation level on the first interface according to a preset display method for the target type.

[0165] According to an embodiment of the present disclosure, the first determination module may include a comparison unit and a first determination unit.

[0166] The comparison unit is configured to compare the information to be evaluated with an index range determined based on historical information to obtain a deviation value, where the index range is associated with the attribute information of the first object, and the deviation value characterizes the deviation degree of the information to be evaluated relative to the index range.

[0167] The first determination unit is configured to determine the evaluation level according to the target type and the deviation value.

[0168] According to an embodiment of the present disclosure, the large model-based information evaluation device 800 may further include a first push module.

[0169] The first push module is configured to, in response to the evaluation level conforming to a predetermined response rule, push a warning message to a second object based on a push channel corresponding to a target type, where the predetermined response rule defines candidate channels and candidate objects for different evaluation levels of each candidate type.

[0170] According to an embodiment of the present disclosure, the information evaluation result includes an evaluation description for the information to be evaluated.

[0171] According to an embodiment of the present disclosure, the information evaluation module 820 may include a processing unit.

[0172] The processing unit is configured to use an information generation large model to process a first prompt message for a target type and the information to be evaluated to obtain an evaluation description, where the first prompt message is used to guide the information generation large model to analyze the information to be evaluated to obtain an evaluation description.

[0173] According to an embodiment of the present disclosure, the predetermined response rule further defines candidate templates for each candidate type.

[0174] According to an embodiment of the present disclosure, the information evaluation module 820 may include a providing unit.

[0175] The providing unit is configured to, based on a second prompt message generated according to a candidate template for a target type, use the information generation large model to perform multiple rounds of interactive conversations with a first object so that the first object provides supplementary information.

[0176] According to an embodiment of the present disclosure, the processing unit includes a processing subunit.

[0177] The processing subunit is configured to use an information generation large model to process the supplementary information, the first prompt message, and the information to be evaluated obtained through multiple rounds of interactive conversations to obtain an evaluation description.

[0178] According to an embodiment of the present disclosure, the large model-based information evaluation device 800 may include an acquisition module and a first storage module.

[0179] The acquisition module is configured to, in response to receiving a trigger message, use a third interface to acquire attribute information and auxiliary information for a candidate object, where the auxiliary information includes at least one of the following: visual information and voiceprint information.

[0180] The first storage module is configured to associatively store the candidate identifier, visual information, and auxiliary information of the candidate object in an object map.

[0181] According to an embodiment of the present disclosure, the object mapping includes the correspondence relationships among multiple groups of candidate identifiers, attribute information, and auxiliary information, and the evaluation request includes input information, where the input information includes at least one of the second image of the first object and the audio to be evaluated of the first object.

[0182] According to an embodiment of the present disclosure, the large model-based information evaluation device 800 may include a second determination module and a second storage module.

[0183] The second determination module is configured to determine the auxiliary information for the first object according to the object identifier of the first object and the object mapping.

[0184] The second storage module is configured to update the information to be evaluated and the information evaluation result as the historical information of the first object to the object mapping in response to the input information matching the auxiliary information.

[0185] According to an embodiment of the present disclosure, the large model-based information evaluation device 800 may include a generation module and a second push module.

[0186] The generation module is configured to generate push information at intervals of a predetermined time period according to the historical information corresponding to the first object in the object mapping in response to the first object having an associated object.

[0187] The second push module is configured to push the push information to the associated object based on the push channel corresponding to the associated object.

[0188] Figure 9 The structural block diagram of the agent of the large model according to an embodiment of the present disclosure is schematically shown.

[0189] In an embodiment of the present disclosure, inspired by the von Neumann architecture in modern computer theory, as Figure 9 shown, the AI agent 900 may include five core modules: an input module 910, a control module 920, a storage module 930, an operation module 940, and an output module 950.

[0190] The input module 910 is responsible for receiving or perceiving information such as queries, requests, instructions, signals, or data from the outside world (such as users or the external environment), and converting it into a format that the AI agent 900 can understand and process. The input module 910 is the primary link for the AI agent 900 to interact with the outside world. It enables the AI agent 900 to efficiently and accurately obtain the necessary "sensory" information from the outside world and respond to this information.

[0191] In the example, the input module 910 may input the evaluation request described above.

[0192] The control module 920 is the core support for the AI agent 900 to handle complex tasks. During the model training phase, the control module 920 can execute the information evaluation method based on the large model described above.

[0193] In the example, during operation, the control module 920 will continuously interact with the storage module 930, the operation module 940, and / or the output module 950. However, it should be noted that in the embodiments of the present disclosure, the control module 920 acts as a single initiator to initiate communication with the storage module 930, the operation module 940, and / or the output module 950, and there is no communication coupling between the storage module 930, the operation module 940, and the output module 950.

[0194] In the example, the performance of the control module 920 can be closely related to the large model on which the AI agent 900 is based. To fully utilize the capabilities of the large language model, the internal structure of the control module 920 can be designed to be highly configurable and extensible to handle various different types of tasks and requirements in real-world scenarios.

[0195] The storage module 930 can be responsible for memorizing the trained video generation model. Various types of historical information of the first object as described above can be included in the storage module 930.

[0196] In the example, after the AI agent 900 receives an evaluation request, the AI agent 900 can trigger the information evaluation process based on the large model, obtain the historical information of the first object with the target type from the storage module 930, and feedback it to the control module 920. Then, the control module 920 can transfer the feedback historical information of the first object with the target type to the output module 950.

[0197] The operation module 940 can be regarded as a predefined tool library. Tools for device classification and tools for text recognition as described above can be included in the operation module 940.

[0198] In the example, when the AI agent 900 needs to process data, relevant tools can be called from the operation module 940 and feedback to the control module 920. Then, the control module 920 can use the feedback tools to process the first image to obtain the information to be evaluated. It can be understood that although the large language model has excellent language understanding and generation capabilities, like humans, the tasks it can solve without any tools are very limited. When the AI agent 900 is given the ability to call tools, tasks such as completing device classification with the help of tools for device classification can be achieved.

[0199] During the model training phase, the output module 950 can output the information evaluation result described above.

[0200] The AI agent 900 according to the embodiments of the present disclosure can simply and effectively improve the degree of intelligence, and improve flexibility and versatility.

[0201] Figure 10 A block diagram of an electronic device suitable for implementing an information evaluation method based on a large model according to an embodiment of the present disclosure is schematically shown. The electronic device is intended to represent various forms of digital computers, such as, for example, a laptop computer, a desktop computer, a workbench, a personal digital assistant, a server, a blade server, a mainframe computer, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as, for example, a personal digital processor, a cellular phone, a smart phone, a wearable device, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely exemplary and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0202] As Figure 10 shown, the device 1000 includes a computing unit 1001, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 1002 or a computer program loaded from a storage unit 1008 into a random access memory (RAM) 1003. In the RAM 1003, various programs and data required for the operation of the device 1000 can also be stored. The computing unit 1001, the ROM 1002, and the RAM 1003 are connected to each other through a bus 1004. An input / output (I / O) interface 1005 is also connected to the bus 1004.

[0203] A plurality of components in the device 1000 are connected to the I / O interface 1005, including: an input unit 1006, such as a keyboard, a mouse, etc.; an output unit 1007, such as various types of displays, speakers, etc.; a storage unit 1008, such as a magnetic disk, an optical disk, etc.; and a communication unit 1009, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 1009 allows the device 1000 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0204] The computing unit 1001 can be various general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of the computing unit 1001 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 1001 executes the various methods and processes described above, such as the information evaluation method based on a large model. For example, in some embodiments, the information evaluation method based on a large model can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 1008. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 1000 via the ROM 1002 and / or the communication unit 1009. When the computer program is loaded into the RAM 1003 and executed by the computing unit 1001, one or more steps of the information evaluation method based on a large model described above can be executed. Alternatively, in other embodiments, the computing unit 1001 can be configured to execute the information evaluation method based on a large model in any other suitable manner (e.g., by means of firmware).

[0205] Various embodiments of the systems and techniques described above in this document can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a dedicated or general-purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0206] The program code for implementing the methods of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to the processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when the program code is executed by the processor or controller, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The program code can be executed entirely on the machine, partially on the machine, as an independent software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0207] In the context of this disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in connection with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber, a portable compact disc read-only memory (CDROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0208] In order to provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, speech input, or tactile input).

[0209] The systems and techniques described herein can be implemented in a computing system including backend components (e.g., as a data server), or a computing system including middleware components (e.g., an application server), or a computing system including frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system including any combination of such backend components, middleware components, or frontend components. The components of the system can be interconnected to each other by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), and the Internet.

[0210] A computer system can include a client and a server. The client and the server are generally remote from each other and typically interact through a communication network. The client-server relationship is generated by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, a server of a distributed system, or a server incorporating a blockchain.

[0211] It should be understood that the various forms of processes shown above can be used, with steps reordered, added or deleted. For example, the steps described in the present disclosure can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution disclosed in the present disclosure can be achieved, and no limitation is imposed herein.

[0212] The above specific embodiments do not constitute a limitation on the protection scope of the present disclosure. Those skilled in the art should understand that various modifications, combinations, sub - combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present disclosure shall be included within the protection scope of the present disclosure.

Claims

1. An information evaluation method based on a large model, comprising: In response to receiving an evaluation request for a first object, by recognizing a first image indicated by the evaluation request, to display information to be evaluated on a first interface, wherein the first image is a display interface image including a target device for collecting information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated; and Inputting the information to be evaluated and historical information of the first object having the target type into an information generation large model to display an information evaluation result on the first interface, wherein the information evaluation result is determined according to the relationship between the information to be evaluated and the historical information.

2. The method according to claim 1, wherein, The step of recognizing the first image indicated by the evaluation request to display the information to be evaluated on the first interface includes: Classifying the target device in the first image according to the morphological characteristics of each of a plurality of candidate devices to obtain the target type; and According to the target type, performing text recognition on the first image to obtain and display the information to be evaluated on the first interface.

3. The method according to claim 2, wherein The step of performing text recognition on the first image according to the target type to obtain and display the information to be evaluated on the first interface includes: Intercepting the information to be evaluated in the display interface image according to the target type to obtain a region image, wherein the region image is a partial image of the first image that displays the information to be evaluated; and Performing text recognition on the region image to obtain the information to be evaluated.

4. The method according to claim 1, wherein The step of recognizing the first image indicated by the evaluation request to display the information to be evaluated on the first interface includes: Using a pre-trained recognition model to recognize the first image to obtain and display the information to be evaluated on the first interface, wherein the recognition model is trained using training samples, and the training samples include images of each of a plurality of candidate devices under the influence of different perspective parameters and different environmental parameters; Wherein the recognition model includes a device recognition network and a text recognition network, the device recognition network is used to classify the target device in the first image according to the morphological characteristics of each of a plurality of candidate devices to obtain the target type, and the text recognition network is used to perform text recognition on the first image according to the target type to obtain the information to be evaluated.

5. The method according to claim 4, wherein, The training samples include a plurality of positive sample pairs and a plurality of negative sample pairs, the positive sample pairs and the negative sample pairs are dynamically updated at intervals of a predetermined time period, and the number of positive sample pairs and the number of negative sample pairs satisfy a predetermined balance condition. The sample images included in the positive sample pairs match the label type of the candidate device, and the sample images included in the negative sample pairs do not match the label type.

6. The method according to claim 1, wherein, The information evaluation result includes the evaluation level to which the information to be evaluated belongs; The method further includes: Based on the target type, determining the evaluation level according to the relationship between the information to be evaluated and the historical information; and Display the evaluation level on the first interface according to a preset display method for the target type.

7. The method according to claim 6, wherein, The determining the evaluation level based on the target type and according to the relationship between the information to be evaluated and the historical information includes: Comparing the information to be evaluated with an index range determined based on the historical information to obtain a deviation value, where the index range is associated with the attribute information of the first object, and the deviation value characterizes the degree of deviation of the information to be evaluated relative to the index range; and Determining the evaluation level according to the target type and the deviation value.

8. The method according to claim 6, further comprising: In response to the evaluation level conforming to a predetermined response rule, pushing a warning message to a second object based on a push channel corresponding to the target type, where the predetermined response rule defines candidate channels and candidate objects for different evaluation levels of each candidate type.

9. The method according to claim 8, wherein The information evaluation result includes an evaluation description for the information to be evaluated; The inputting the information to be evaluated and the historical information of the first object having the target type into an information generation large model to display an information evaluation result on the first interface includes: Using the information generation large model to process a first prompt information for the target type and the information to be evaluated to obtain the evaluation description, where the first prompt information is used to guide the information generation large model to analyze the information to be evaluated to obtain the evaluation description.

10. The method according to claim 9, wherein, The predetermined response rule further defines candidate templates for each of the candidate types; The method further comprises: Based on a second prompt information generated according to a candidate template for the target type, using the information generation large model to conduct multiple rounds of interactive conversations with the first object to enable the first object to provide supplementary information; The using the information generation large model to process the first prompt information and the information to be evaluated to obtain the evaluation description includes: Using the information generation large model to process the supplementary information, the first prompt information, and the information to be evaluated obtained through multiple rounds of interactive conversations to obtain the evaluation description.

11. The method according to claim 1, further comprising: In response to receiving a trigger information, using a third interface to obtain the attribute information and auxiliary information for a candidate object, where the auxiliary information includes at least one of the following: visual information and voiceprint information; and Associatively storing the candidate identifier, the visual information, and the auxiliary information of the candidate object in an object mapping.

12. The method according to claim 11, wherein, The object mapping includes multiple groups of corresponding relationships between the candidate identifier, the attribute information, and the auxiliary information, the evaluation request includes input information, and the input information includes at least one of a second image of the first object and the audio to be evaluated of the first object; The method further comprises: Determining the auxiliary information for the first object according to the object identifier of the first object and the object mapping; And In response to the input information matching the auxiliary information, update the information to be evaluated and the information evaluation result as historical information of the first object to the object mapping.

13. The method according to claim 11, further comprising: In response to the first object having an associated object, at intervals of a predetermined time period, generate push information according to the historical information corresponding to the first object in the object mapping; and Based on the push channel corresponding to the associated object, push the push information to the associated object.

14. An information evaluation device based on a large model, comprising: An image recognition module, configured to, in response to receiving an evaluation request for a first object, identify a first image indicated by the evaluation request to display information to be evaluated on a first interface, where the first image is a display interface image including a target device for collecting information to be evaluated of the first object having a target type, and the display interface image includes the information to be evaluated; and An information evaluation module, configured to input the information to be evaluated and the historical information of the first object having the target type into an information generation large model to display an information evaluation result on the first interface, where the information evaluation result is determined according to the relationship between the information to be evaluated and the historical information.

15. An intelligent agent of artificial intelligence, configured to execute the method according to any one of claims 1 to 13.

16. An electronic device, comprising: One or more processors; A memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 13.

17. A computer-readable storage medium having a computer program or instructions stored thereon, characterized in that, The computer program or instruction, when executed by the processor, implements the steps of the method according to any one of claims 1 to 13.

18. A computer program product, comprising a computer program or instructions, characterized in that, The computer program or instruction, when executed by the processor, implements the steps of the method according to any one of claims 1 to 13.

Images