Intelligent medicine classification system

Abstract

The invention relates to an intelligent medicine classification system, and relates to the technical field of artificial intelligence. The system mainly comprises the following steps: acquiring an outer package image of a to-be-classified medicine through a graph acquisition device; determining whether a medicine identifier uniquely corresponding to the to-be-classified medicine can be obtained or not according to the outer package image; if the drug identifier uniquely corresponding to the to-be-classified drug can be obtained, obtaining drug detailed data corresponding to the drug identifier according to the drug database; if the drug identifier uniquely corresponding to the to-be-classified drug cannot be acquired, acquiring a specification image of the to-be-classified drug; determining corresponding detailed medicine data according to the specification image; and processing the detailed data of the medicine based on a classification basis defined by a user to obtain a classification result corresponding to the to-be-classified medicine.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent drug classification system. Background Art

[0002] At present, there is no unified standard for drug classification and representative product selection in the industry. At the same time, due to the large number of drug categories, manual review and classification is time-consuming and labor-intensive, and it is difficult to ensure the systematicness and integrity of the classification. The classification of drugs and the selection of representative products are crucial to payment standards, affecting the prescription choices of clinicians and the medication choices of patients, and thus affecting their effective value in clinical practice. Summary of the invention

[0003] The present invention is intended to provide an intelligent drug classification system to solve the deficiencies in the prior art. The technical problem to be solved by the present invention is achieved through the following technical solutions.

[0004] An embodiment of the present invention provides an intelligent drug classification system, the system comprising: an image acquisition device, a drug database, and a drug classification device, the drug classification device comprising: a memory and at least one processor, the memory storing a computer program, the at least one processor being configured to implement the following steps when executing the computer program:

[0005] Acquire the outer packaging image of the drug to be classified by the image acquisition device;

[0006] Determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image;

[0007] If a drug identification uniquely corresponding to the drug to be classified can be obtained, then obtaining drug detailed data corresponding to the drug identification according to the drug database;

[0008] If the drug identification uniquely corresponding to the drug to be classified cannot be obtained, then obtaining the instruction image of the drug to be classified;

[0009] Determine the corresponding detailed data of the drug according to the instruction image;

[0010] The detailed data of the drug is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified.

[0011] In an optional embodiment, determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image includes:

[0012] Inputting the outer packaging image into an image feature recognition model to obtain image recognition features corresponding to the outer packaging image; the image recognition features at least include color features, texture features, graphic structure features and edge features;

[0013] By calculating the similarity between the image recognition feature and the drug image feature in the drug database, it is determined whether a drug identification uniquely corresponding to the drug to be classified can be obtained.

[0014] In an optional embodiment, determining whether a drug identifier uniquely corresponding to the drug to be classified can be obtained by calculating the similarity between the image recognition feature and the drug image feature in the drug database includes:

[0015] Calculate the similarity between the color features, texture features, graphic structure features and edge features in the image recognition features and the color features, texture features, graphic structure features and edge features in each drug image feature in the drug database to obtain color similarity, texture similarity, graphic structure similarity and edge similarity;

[0016] Performing weighted calculation on the color similarity, the texture similarity, the graphic structure similarity and the edge similarity to obtain the similarity of each drug image feature in the drug database;

[0017] The drug identification corresponding to the drug image feature in the drug database whose similarity exceeds a preset value and has the highest similarity is determined as the drug identification uniquely corresponding to the drug to be classified.

[0018] In an optional embodiment, the step of inputting the outer packaging image into an image feature recognition model to obtain the image recognition feature corresponding to the outer packaging image includes:

[0019] Inputting the outer packaging image into an image feature recognition model, and obtaining initial image data features through a feature extraction layer in the image feature recognition model;

[0020] The initial image data features are respectively input into a color feature extraction module, a texture feature extraction module, a graphic structure extraction module, and an edge detection module to obtain color features, texture features, graphic structure features, and edge features corresponding to the outer packaging image.

[0021] In an optional embodiment, determining corresponding detailed drug data according to the instruction sheet image includes:

[0022] Performing image recognition on the instruction sheet image by using image recognition technology to obtain initial drug instruction data;

[0023] Inputting the instruction image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data;

[0024] The detailed drug data corresponding to the instruction sheet image is determined according to the initial drug instruction data and the intermediate drug instruction data.

[0025] In an optional embodiment, the determining the detailed drug data corresponding to the instruction sheet image according to the initial drug instruction data and the intermediate drug instruction data includes:

[0026] determining a data difference position between the initial drug description data and the intermediate drug description data;

[0027] Performing text analysis on the context of the data distinguishing position to determine the final data text content of the data distinguishing position to obtain final intermediate drug description data;

[0028] The final intermediate drug description data is determined as the drug detailed data corresponding to the instruction manual image.

[0029] In an optional embodiment, the step of inputting the instruction image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data includes:

[0030] Inputting the instruction manual image and the initial drug data into a text recognition model, and obtaining the instruction manual image vector and the initial drug data vector through a convolutional layer in the text recognition model;

[0031] Input the instruction manual image vector into a first prediction module to obtain a first prediction text result; input the initial drug data vector into a second prediction module to obtain a second prediction text result;

[0032] By comparing the first predicted text result and the second predicted text result, the intermediate drug description data corresponding to the initial drug data is determined.

[0033] In an optional embodiment, the detailed drug data is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified, including:

[0034] Obtaining a classification basis defined by a user, and determining a corresponding data field according to the classification basis;

[0035] Extracting text from the detailed drug data according to the data field to obtain corresponding data content;

[0036] The classification result corresponding to the drug to be classified is determined according to the data content.

[0037] In an optional embodiment, extracting text from the drug detailed data according to the data field to obtain corresponding data content includes:

[0038] Determine a field set corresponding to the data field, where the fields in the field set are synonyms and / or near synonyms of the data field;

[0039] Matching the field set with the drug detailed data to determine the corresponding data field position in the drug detailed data;

[0040] The data content related to the data field position is extracted from the drug detailed data.

[0041] An embodiment of the present invention provides an intelligent drug classification method, the method comprising:

[0042] Acquire the outer packaging image of the drug to be classified by means of a graphics acquisition device;

[0043] Determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image;

[0044] If a drug identification uniquely corresponding to the drug to be classified can be obtained, then obtaining detailed drug data corresponding to the drug identification according to the drug database;

[0045] If the drug identification uniquely corresponding to the drug to be classified cannot be obtained, then obtaining the instruction image of the drug to be classified;

[0046] Determine the corresponding detailed data of the drug according to the instruction image;

[0047] The detailed drug data is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified.

[0048] The embodiments of the present invention include the following advantages:

[0049] An intelligent drug classification system provided by an embodiment of the present invention comprises: an image acquisition device, a drug database and a drug classification device, wherein the drug classification device comprises: a memory and at least one processor, wherein the memory stores a computer program, and the at least one processor is configured to implement the following steps when executing the computer program: first, obtaining an outer packaging image of the drug to be classified through a graphic acquisition device; then determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image; if a drug identification uniquely corresponding to the drug to be classified can be obtained, obtaining drug detailed data corresponding to the drug identification according to the drug database; if a drug identification uniquely corresponding to the drug to be classified cannot be obtained, obtaining an instruction image of the drug to be classified; determining the corresponding drug detailed data according to the instruction image; processing the drug detailed data based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified. Compared with the current manual review of drug classification, the present application automatically determines the classification result of the drug to be classified based on the outer packaging image of the drug to be classified, thereby improving the efficiency and accuracy of drug classification through the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a flow chart of an intelligent drug classification method provided by an embodiment of the present invention;

[0051] Figure 2 is a flowchart for determining detailed drug data provided by an embodiment of the present invention;

[0052] Figure 3 It is a structural schematic diagram of an intelligent drug classification device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0053] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0054] An embodiment of the present invention provides an intelligent drug classification system, which includes: an image acquisition device, a drug database, and a drug classification device, wherein the drug classification device includes: a memory and at least one processor, wherein the memory stores a computer program. Figure 1 The flowchart of a smart drug classification method provided in this embodiment, at least one processor is configured to implement the following steps when executing the computer program:

[0055] S101, obtaining an image of the outer packaging of the drug to be classified through the image acquisition device.

[0056] In this embodiment, the image acquisition device can be a device that can capture images, such as a mobile phone, a handheld scanner, etc. Specifically, in this embodiment, the outer packaging of the medicine to be classified can be captured by a mobile phone to obtain an outer packaging image, so as to classify the medicine to be classified according to the outer packaging image.

[0057] It should be noted that this embodiment can capture images of any one or more sides of the outer packaging of the drug to be classified. Preferably, this embodiment can capture a front image (an image with obvious features) of the drug to be classified, and then use the captured front image as the outer packaging image of the drug to be classified.

[0058] S102: Determine whether a drug identification uniquely corresponding to the drug to be classified can be obtained based on the outer packaging image.

[0059] Among them, the drug identification can uniquely identify the corresponding drug to be classified. The drug identification can be composed of numbers and letters. The corresponding drug to be classified can be uniquely determined through the drug identification.

[0060] In an optional embodiment provided in the present application, determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image includes:

[0061] S1021: Input the outer packaging image into an image feature recognition model to obtain image recognition features corresponding to the outer packaging image.

[0062] The image recognition features include at least color features, texture features, graphic structure features and edge features.

[0063] Specifically, the step of inputting the outer packaging image into an image feature recognition model to obtain image recognition features corresponding to the outer packaging image includes: inputting the outer packaging image into the image feature recognition model, and obtaining initial image data features through a feature extraction layer in the image feature recognition model; and inputting the initial image data features into a color feature extraction module, a texture feature extraction module, a graphic structure extraction module, and an edge detection module, respectively, to obtain color features, texture features, graphic structure features, and edge features corresponding to the outer packaging image.

[0064] Among them, the color feature extraction module is used to extract the surface properties corresponding to the image or image area. It is based on the characteristics of pixel points and reflects the global distribution of colors in the image, such as color histogram, color set, color moment, etc.; the texture feature extraction module is used to extract the surface properties corresponding to the image or image area, but unlike color features, texture features are not based on pixel features, but need to be statistically calculated in an area containing multiple pixels; the graphic structure extraction module is used for the outer boundary of the object, while the regional features of the image are related to the entire shape area, and the mutual spatial position or relative direction relationship between multiple targets segmented from the image. These relationships can also be divided into connection / adjacency relationships, overlap / overlapping relationships and inclusion / inclusion relationships, etc.; the edge detection module is used to extract the locations where the brightness changes sharply in the image, usually using edge detection algorithms (such as Canny, Sobel, Laplace, etc.) to extract.

[0065] In this embodiment, the feature extraction layer in the image feature recognition model is used to extract the initial image data features in the image, and the feature extraction model can adopt the existing model structure. In the process of training the image feature recognition model, the sample image needs to be input into the image feature recognition model, and then the initial image data features corresponding to the sample image are obtained, and then the initial image data features are respectively input into the color feature extraction module, the texture feature extraction module, the graphic structure extraction module, and the edge detection module to obtain the predicted color features, texture features, graphic structure features, and edge features, and then the loss value of the image feature recognition model is calculated according to the actual color features, texture features, graphic structure features, and edge features of the sample image. When the loss value is less than a certain value, the training of the image feature recognition model is completed.

[0066] Among them, the actual color features of the sample image can be determined according to color histogram, color set, color moment, etc.; the actual texture features and graphic structure features of the sample image can be determined by a convolutional network model; the actual edge features of the sample image can be determined by an edge detection algorithm, and this embodiment does not make any specific limitations on this.

[0067] S1022, determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained by calculating the similarity between the image recognition feature and the drug image feature in the drug database.

[0068] The drug database stores drug image features and drug identifications corresponding to a plurality of drugs. In this embodiment, the drug identification corresponding to the drug to be classified can be determined by calculating the similarity between the image recognition feature and the drug image feature in the drug database, that is, when the image recognition feature matches the drug image feature in the drug database successfully, the drug identification corresponding to the successfully matched drug image feature is determined as the drug identification corresponding to the drug to be classified. In this embodiment, the drug image feature is successfully matched, which can be that the similarity exceeds a preset value, or the similarity is the highest.

[0069] Specifically, the method of determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained by calculating the similarity between the image recognition feature and the drug image feature in the drug database includes: respectively calculating the similarity between the color feature, texture feature, graphic structure feature and edge feature in the image recognition feature and the color feature, texture feature, graphic structure feature and edge feature in each drug image feature in the drug database to obtain color similarity, texture similarity, graphic structure similarity and edge similarity; performing weighted calculation on the color similarity, texture similarity, graphic structure similarity and edge similarity to obtain the similarity of each drug image feature in the drug database; determining the drug identification corresponding to the drug image feature in the drug database whose similarity exceeds a preset value and has the highest similarity as the drug identification uniquely corresponding to the drug to be classified. Among them, this embodiment can use various distance measurement methods to calculate the similarity of drug image features, such as Euclidean distance, cosine similarity, Hamming distance, etc.

[0070] S103: If a drug identification uniquely corresponding to the drug to be classified can be obtained, then the drug detailed data corresponding to the drug identification is obtained according to the drug database.

[0071] Among them, in addition to storing the drug image features and drug identifications corresponding to a variety of drugs, the drug database also stores the corresponding drug detailed data, through which the classification of drugs to be classified can be realized.

[0072] S104: If the drug identification uniquely corresponding to the drug to be classified cannot be obtained, the instruction sheet image of the drug to be classified is obtained.

[0073] The instruction image of the drug to be classified may be an instruction image built into the drug packaging, or may be an outer packaging image of other surfaces of the drug to be classified, and this embodiment does not make any specific limitation to this.

[0074] S105, determining corresponding detailed drug data according to the instruction manual image.

[0075] like Figure 2The figure shows a flow chart of determining detailed drug data provided by this embodiment. In this embodiment, determining the corresponding detailed drug data according to the instruction image includes:

[0076] S1051, performing image recognition on the instruction manual image by using image recognition technology to obtain initial drug instruction data.

[0077] Specifically, this embodiment can first preprocess the input instruction manual image to improve the quality of the instruction manual image and reduce the difficulty of subsequent processing. Preprocessing includes operations such as denoising, binarization, and grayscale. Then use image processing technology (such as edge detection and contour analysis) to locate the text area in the image. After that, the text in the text area is divided into single characters or words, and features are extracted for each character, such as shape, angle, texture, etc. The extracted character features are compared with the pre-trained model to identify the corresponding characters, thereby obtaining the initial drug description data corresponding to the instruction manual image.

[0078] S1052, inputting the instruction manual image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data.

[0079] Specifically, the step of inputting the instruction manual image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data includes: inputting the instruction manual image and the initial drug data into a text recognition model, and obtaining an instruction manual image vector and an initial drug data vector through a convolutional layer in the text recognition model; inputting the instruction manual image vector into a first prediction module to obtain a first predicted text result; inputting the initial drug data vector into a second prediction module to obtain a second predicted text result; and determining the intermediate drug instruction data corresponding to the initial drug data by comparing the first predicted text result with the second predicted text result.

[0080] In this embodiment, the first predicted text result and the second predicted text result are compared to determine whether the two results are consistent. If they are consistent, the result is directly used as the intermediate drug description data corresponding to the initial drug data; if they are inconsistent, the specific content of the location where the content in the predicted text result is different is determined through context analysis, and then the re-determined predicted text result is determined as the intermediate drug description data corresponding to the initial drug description data.

[0081] Preferably, the present embodiment can also determine the intermediate drug description data corresponding to the initial drug data according to the weight values ​​of the first prediction module and the second prediction module. The weight value can be determined according to the accuracy of the first prediction module and the second prediction module, and the higher the accuracy, the larger the corresponding weight value.

[0082] In this embodiment, the training process of the text recognition model may be:

[0083] 1. Obtain sample data (sample data includes sample images and sample drug data) and corresponding actual drug description data; the sample drug data is obtained in the same manner as in step S1051;

[0084] 2. Input the sample image and sample drug data into the text recognition model, obtain the first prediction result based on the sample image, and obtain the second prediction result based on the sample drug data;

[0085] 3. Calculate the loss value based on the first prediction result, the second prediction result and the actual drug description data; the calculation formula of the loss value is:

[0086]

[0087] Among them, n is the number of sample data, x i is the first prediction result in the i-th sample data, y i is the second prediction result in the i-th sample data, z i is the actual drug description data corresponding to the i-th sample data, (x i , z i ) is the similarity between the first prediction result and the actual drug description data, (y i , z i ) is the similarity between the second prediction result and the actual drug description data.

[0088] 4. When the loss value is less than the predetermined value, the training of the text recognition model is completed.

[0089] S1053, determining the detailed drug data corresponding to the instruction sheet image according to the initial drug description data and the intermediate drug description data.

[0090] Specifically, determining the detailed drug data corresponding to the instruction manual image based on the initial drug description data and the intermediate drug description data includes: determining the data distinction position between the initial drug description data and the intermediate drug description data; performing text analysis on the context of the data distinction position to determine the final data text content of the data distinction position to obtain the final intermediate drug description data; and determining the final intermediate drug description data as the detailed drug data corresponding to the instruction manual image.

[0091] S106, processing the detailed drug data based on the classification basis defined by the user to obtain a classification result corresponding to the drug to be classified.

[0092] In this embodiment, drugs can be classified based on the safety and effectiveness principles of drugs, as well as different characteristics of their varieties, specifications, indications, dosages, and routes of administration. For example, drugs can be classified into prescription drugs and over-the-counter drugs based on the safety and effectiveness principles of drugs, classified according to their functions and indications, or classified according to their sites of action, therapeutic uses, and chemical structures.

[0093] In addition, this application can also classify scarce drugs according to the five characteristics of shortage drugs, namely, existing in the market, clinically necessary, irreplaceable, high price and small quantity, and raw material dependence, as follows:

[0094] Category I, emergency level 0: drugs that are already available in the market but are not clinically necessary.

[0095] The second category, emergency level 2: drugs that are already available in the market, clinically necessary and can be replaced.

[0096] Category III, emergency level 2: drugs that already exist in the market, are clinically necessary, irreplaceable, and available in large quantities.

[0097] Category 4, level 3 urgency: drugs that already exist in the market, are clinically necessary, irreplaceable, and are high in price but small in quantity.

[0098] Category 5, level 4 urgency: drugs that already exist in the market, are clinically necessary, irreplaceable, high in price and small in quantity, and dependent on raw materials.

[0099] It should be noted that the grading and classification of scarce drugs in this embodiment requires not only detailed drug data, but also market data, clinical data, sales data, etc. corresponding to the drug, so as to achieve the grading and classification of scarce drugs.

[0100] In an optional embodiment provided in the present application, the detailed drug data is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified, including:

[0101] S1061, obtaining a classification basis defined by the user, and determining a corresponding data field according to the classification basis.

[0102] S1062: Extract text from the drug detailed data according to the data field to obtain corresponding data content.

[0103] Specifically, the text extraction from the drug detailed data according to the data field to obtain the corresponding data content includes: determining a field set corresponding to the data field, the fields in the field set being synonyms and / or near-synonyms of the data field; matching the field set with the drug detailed data to determine the corresponding data field position in the drug detailed data; and extracting data content related to the data field position from the drug detailed data.

[0104] S1063, determining the classification result corresponding to the drug to be classified according to the data content.

[0105] This embodiment provides an intelligent drug classification method, firstly, the outer packaging image of the drug to be classified is obtained through a graphic acquisition device; then, it is determined whether the drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image; if the drug identification uniquely corresponding to the drug to be classified can be obtained, the detailed drug data corresponding to the drug identification is obtained according to the drug database; if the drug identification uniquely corresponding to the drug to be classified cannot be obtained, the instruction image of the drug to be classified is obtained; the corresponding detailed drug data is determined according to the instruction image; the detailed drug data is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified. Compared with the current manual review and classification of drugs, the present application automatically determines the classification result of the drug to be classified based on the outer packaging image of the drug to be classified, so that the efficiency and accuracy of drug classification can be improved through the present application.

[0106] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0107] In one embodiment, a smart drug classification device is provided. Figure 3 As shown, the functional modules of the intelligent drug classification device are described in detail as follows:

[0108] An acquisition module 31 is used to acquire an image of the outer packaging of the drug to be classified through a graphic acquisition device;

[0109] A determination module 32, configured to determine whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image;

[0110] The acquisition module 31 is further configured to acquire detailed drug data corresponding to the drug identification according to the drug database if a drug identification uniquely corresponding to the drug to be classified can be acquired;

[0111] The acquisition module 31 is further configured to acquire the instruction sheet image of the drug to be classified if the drug identification uniquely corresponding to the drug to be classified cannot be acquired;

[0112] The determination module 32 is further used to determine the corresponding drug detailed data according to the instruction image;

[0113] The classification module 33 is used to process the detailed data of the drug based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified.

[0114] In an optional embodiment, the determination module 32 is specifically configured to:

[0115] Inputting the outer packaging image into an image feature recognition model to obtain image recognition features corresponding to the outer packaging image; the image recognition features at least include color features, texture features, graphic structure features and edge features;

[0116] By calculating the similarity between the image recognition feature and the drug image feature in the drug database, it is determined whether a drug identification uniquely corresponding to the drug to be classified can be obtained.

[0117] In an optional embodiment, the determination module 32 is specifically configured to:

[0118] Calculate the similarity between the color features, texture features, graphic structure features and edge features in the image recognition features and the color features, texture features, graphic structure features and edge features in each drug image feature in the drug database to obtain color similarity, texture similarity, graphic structure similarity and edge similarity;

[0119] Performing weighted calculation on the color similarity, the texture similarity, the graphic structure similarity and the edge similarity to obtain the similarity of each drug image feature in the drug database;

[0120] The drug identification corresponding to the drug image feature in the drug database whose similarity exceeds a preset value and has the highest similarity is determined as the drug identification uniquely corresponding to the drug to be classified.

[0121] In an optional embodiment, the determination module 32 is specifically configured to:

[0122] Inputting the outer packaging image into an image feature recognition model, and obtaining initial image data features through a feature extraction layer in the image feature recognition model;

[0123] The initial image data features are respectively input into a color feature extraction module, a texture feature extraction module, a graphic structure extraction module, and an edge detection module to obtain color features, texture features, graphic structure features, and edge features corresponding to the outer packaging image.

[0124] In an optional embodiment, the determination module 32 is specifically configured to:

[0125] Performing image recognition on the instruction sheet image by using image recognition technology to obtain initial drug instruction data;

[0126] Inputting the instruction image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data;

[0127] The detailed drug data corresponding to the instruction sheet image is determined according to the initial drug instruction data and the intermediate drug instruction data.

[0128] In an optional embodiment, the determination module 32 is specifically configured to:

[0129] determining a data difference position between the initial drug description data and the intermediate drug description data;

[0130] Performing text analysis on the context of the data distinguishing position to determine the final data text content of the data distinguishing position to obtain final intermediate drug description data;

[0131] The final intermediate drug description data is determined as the drug detailed data corresponding to the instruction manual image.

[0132] In an optional embodiment, the determination module 32 is specifically configured to:

[0133] Inputting the instruction manual image and the initial drug data into a text recognition model, and obtaining the instruction manual image vector and the initial drug data vector through a convolutional layer in the text recognition model;

[0134] Input the instruction manual image vector into a first prediction module to obtain a first prediction text result; input the initial drug data vector into a second prediction module to obtain a second prediction text result;

[0135] By comparing the first predicted text result and the second predicted text result, the intermediate drug description data corresponding to the initial drug data is determined.

[0136] In an optional embodiment, the classification module 33 is specifically used for:

[0137] Obtaining a classification basis defined by a user, and determining a corresponding data field according to the classification basis;

[0138] Extracting text from the detailed drug data according to the data field to obtain corresponding data content;

[0139] The classification result corresponding to the drug to be classified is determined according to the data content.

[0140] In an optional embodiment, the classification module 33 is specifically used for:

[0141] Determine a field set corresponding to the data field, where the fields in the field set are synonyms and / or near synonyms of the data field;

[0142] Matching the field set with the drug detailed data to determine the corresponding data field position in the drug detailed data;

[0143] The data content related to the data field position is extracted from the drug detailed data.

[0144] It should be noted that the above detailed descriptions are exemplary and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art to which the present application belongs.

[0145] For the specific limitations of the intelligent drug classification device, please refer to the limitations of the intelligent drug classification method above, which will not be repeated here. Each module in the above device can be implemented in whole or in part by software, hardware and a combination thereof. The above modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0146] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0147] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.

Claims

1. An intelligent drug classification system, characterized in that: The system comprises: an image acquisition device, a drug database and a drug classification device, wherein the drug classification device comprises: a memory and at least one processor, wherein the memory stores a computer program, and the at least one processor is configured to implement the following steps when executing the computer program: Acquire the outer packaging image of the drug to be classified by the image acquisition device; Determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image; If a drug identification uniquely corresponding to the drug to be classified can be obtained, then obtaining drug detailed data corresponding to the drug identification according to the drug database; If the drug identification uniquely corresponding to the drug to be classified cannot be obtained, then obtaining the instruction image of the drug to be classified; Determine the corresponding detailed data of the drug according to the instruction image; The detailed data of the drug is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified.

2. The system according to claim 1, characterized in that The determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image includes: Inputting the outer packaging image into an image feature recognition model to obtain image recognition features corresponding to the outer packaging image; the image recognition features at least include color features, texture features, graphic structure features and edge features; By calculating the similarity between the image recognition feature and the drug image feature in the drug database, it is determined whether a drug identification uniquely corresponding to the drug to be classified can be obtained.

3. The system according to claim 2, characterized in that The step of calculating the similarity between the image recognition feature and the drug image feature in the drug database to determine whether a drug identification uniquely corresponding to the drug to be classified can be obtained includes: Calculate the similarity between the color features, texture features, graphic structure features and edge features in the image recognition features and the color features, texture features, graphic structure features and edge features in each drug image feature in the drug database to obtain color similarity, texture similarity, graphic structure similarity and edge similarity; Performing weighted calculation on the color similarity, the texture similarity, the graphic structure similarity and the edge similarity to obtain the similarity of each drug image feature in the drug database; The drug identification corresponding to the drug image feature in the drug database whose similarity exceeds a preset value and has the highest similarity is determined as the drug identification uniquely corresponding to the drug to be classified.

4. The system according to claim 2, characterized in that The step of inputting the outer packaging image into an image feature recognition model to obtain image recognition features corresponding to the outer packaging image includes: Inputting the outer packaging image into an image feature recognition model, and obtaining initial image data features through a feature extraction layer in the image feature recognition model; The initial image data features are respectively input into a color feature extraction module, a texture feature extraction module, a graphic structure extraction module, and an edge detection module to obtain color features, texture features, graphic structure features, and edge features corresponding to the outer packaging image.

5. The system according to claim 1, characterized in that The determining of the corresponding detailed drug data according to the instruction sheet image includes: Performing image recognition on the instruction sheet image by using image recognition technology to obtain initial drug instruction data; Inputting the instruction image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data; The detailed drug data corresponding to the instruction sheet image is determined according to the initial drug instruction data and the intermediate drug instruction data.

6. The system according to claim 5, characterized in that The determining of the detailed drug data corresponding to the instruction sheet image according to the initial drug instruction data and the intermediate drug instruction data includes: determining a data difference position between the initial drug description data and the intermediate drug description data; Performing text analysis on the context of the data distinguishing position to determine the final data text content of the data distinguishing position to obtain final intermediate drug description data; The final intermediate drug description data is determined as the drug detailed data corresponding to the instruction manual image.

7. The system according to claim 5, characterized in that The step of inputting the instruction sheet image and the initial drug data into a text recognition model to determine the intermediate drug instruction data corresponding to the initial drug data includes: Inputting the instruction manual image and the initial drug data into a text recognition model, and obtaining the instruction manual image vector and the initial drug data vector through a convolutional layer in the text recognition model; Input the instruction manual image vector into a first prediction module to obtain a first prediction text result; input the initial drug data vector into a second prediction module to obtain a second prediction text result; By comparing the first predicted text result and the second predicted text result, the intermediate drug description data corresponding to the initial drug data is determined.

8. The system according to any one of claims 1 to 7, characterized in that: The detailed drug data is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified, including: Obtaining a classification basis defined by a user, and determining a corresponding data field according to the classification basis; Extracting text from the detailed drug data according to the data field to obtain corresponding data content; The classification result corresponding to the drug to be classified is determined according to the data content.

9. The system according to claim 8, characterized in that The extracting text from the detailed drug data according to the data field to obtain corresponding data content includes: Determine a field set corresponding to the data field, where the fields in the field set are synonyms and / or near synonyms of the data field; Matching the field set with the drug detailed data to determine the corresponding data field position in the drug detailed data; The data content related to the data field position is extracted from the drug detailed data.

10. An intelligent drug classification method, characterized in that: The method comprises: Acquire the outer packaging image of the drug to be classified by means of a graphics acquisition device; Determining whether a drug identification uniquely corresponding to the drug to be classified can be obtained according to the outer packaging image; If a drug identification uniquely corresponding to the drug to be classified can be obtained, then obtaining detailed drug data corresponding to the drug identification according to the drug database; If the drug identification uniquely corresponding to the drug to be classified cannot be obtained, then obtaining the instruction image of the drug to be classified; Determine the corresponding detailed data of the drug according to the instruction image; The detailed drug data is processed based on the classification basis defined by the user to obtain the classification result corresponding to the drug to be classified.

Images