Traditional Chinese medicine decoction piece identification method based on artificial intelligence

Abstract

The invention discloses a traditional Chinese medicine decoction piece identification method based on artificial intelligence. The method comprises the following steps: acquiring an image of a decoction piece to be decocted; the boiled powder decoction piece image is input into a convolutional neural network recognition model for recognition, a recognition result is returned, and the convolutional neural network recognition model is trained in advance; traversing a database, and obtaining data of each data line corresponding to the decoction pieces according to the identification result; and according to the data of each data line of the corresponding decoction powder decoction pieces, comparing the prescription information, and when the name is inconsistent and / or excessive, marking the decoction powder decoction pieces to be decocted before dosing and dispensing. According to the method, the physical condition of the specific decoction pieces during dosing decoction is identified and checked with the information on the prescription, so that the situation of omission or inconsistent medication or excessive medication is avoided, the dispensing accuracy is improved, and the medication safety of a patient is guaranteed.

Description

Technical Field

[0001] The invention relates to the technical field of identification of Chinese medicine decocted and dispersed pieces during the preparation and decoction process, in particular to an identification method of Chinese medicine decocted and dispersed pieces based on artificial intelligence. Background Art

[0002] Chinese herbal medicine pieces are made into coarse granules or powders of 0.2 to 10 mm in size after processing Chinese herbal medicines according to regulations, boiled in water, and the residue is removed to extract juice or taken together with the residue. In the process of adjusting and boiling Chinese herbal medicine pieces, only the packaging or drug storage grid of the Chinese herbal medicine pieces are identified, and then the medicine is picked up and boiled on the medicine line, or the Chinese herbal medicine pieces are manually screened and identified. The work intensity is high, the identification speed is slow, the work efficiency is low, and it is easy to cause omissions. In addition, the physical condition of the pieces before decoction cannot be directly identified and verified. The step of checking before decoction after picking the medicine is the last link to ensure the safety of patients' medication. The above method lacks identification and verification in this link, and there is a risk of omission or wrong medication due to inconsistent medication, which cannot guarantee the accuracy of adjustment and ensure the safety of patients' medication. Summary of the invention

[0003] In view of the lack of identification and verification in the verification link before decocting the medicine after dispensing, there is a risk of omission or mis-dispensing of medicine due to inconsistent medication. The present invention provides a method for identifying decocted and loose Chinese medicine slices based on artificial intelligence. By identifying the physical conditions of the specific decocted and loose Chinese medicine slices during decocting and verifying them with the information on the prescription, omission or mis-dispensing of medicine can be avoided, thereby improving the accuracy of dispensing and ensuring the safety of patients' medication.

[0004] To achieve the above object, the present invention adopts the following technical scheme: a method for identifying Chinese medicine decoction pieces based on artificial intelligence, comprising the following steps:

[0005] Acquire the image of the decoction pieces to be decocted;

[0006] Inputting the image of the boiled and scattered medicinal pieces into a convolutional neural network recognition model for recognition, and returning a recognition result, wherein the convolutional neural network recognition model is pre-trained;

[0007] Traversing the database, and obtaining data of each data row corresponding to the boiled and dispersed medicinal pieces according to the recognition result;

[0008] Compare the data on the prescription with the data in each row of the corresponding decocted herbal slices. If there is any name discrepancy and / or excess, mark it before the decocted herbal slices are administered and dispensed.

[0009] As a further improvement of the present invention, the obtaining of the image of the decoction pieces to be decocted comprises:

[0010] After reading the Chinese medicine information on the prescription, the decoction system controls the medicine conveyor line to pick up the medicine and collects the images of each boiled and loose medicine piece on the medicine conveyor line.

[0011] As a further improvement of the present invention, the convolutional neural network recognition model includes:

[0012] Collect standard sample images of boiled and loose medicinal pieces;

[0013] The characteristic information is formed by calibrating the standard sample image through the properties and identification methods of each boiled and loose medicinal piece;

[0014] Import the characteristic information of each boiled and dispersed herbal medicine piece into a preset machine learning model for training to obtain a convolutional neural network prediction model;

[0015] The convolutional neural network prediction model is evaluated based on the set model evaluation parameter indicators. If the evaluation passes, a convolutional neural network recognition model is obtained.

[0016] As a further improvement of the present invention, the step of inputting the image of the boiled and scattered medicinal pieces into the convolutional neural network recognition model for recognition comprises:

[0017] The image of the boiled and scattered medicinal pieces contains at least one pixel of boiled and scattered medicinal pieces;

[0018] The images of the boiled and scattered medicinal pieces are input into the trained convolutional neural network recognition model, and the recognition results of the names or identification marks of the boiled and scattered medicinal pieces are output.

[0019] As a further improvement of the present invention, the data of each data row corresponding to the boiled and dispersed decoction pieces is obtained, including:

[0020] Obtain the name or identification mark of the corresponding boiled and dispersed medicinal piece from the recognition result, and use the name or identification mark as a query condition to determine whether there is a data row of the name or the identification mark in the database;

[0021] If it does not exist, the identification process is re-executed;

[0022] If it exists, then obtain the data of all data rows of the corresponding boiled and loose medicinal pieces to which the name or identification mark belongs.

[0023] As a further improvement of the present invention, the database is pre-connected with the traceability information of the Chinese medicine boiled and scattered pieces traceability system so that the current database stores all the boiled and scattered pieces information that can currently be prescribed, and the boiled and scattered pieces information includes the information on the species and breeding, production and processing, sales and transportation, warehousing and the corresponding quality standards of the boiled and scattered pieces of Chinese medicine;

[0024] According to each item of information of the boiled and dispersed medicinal slices, corresponding data items and identification marks for searching or indexing are formed.

[0025] As a further improvement of the present invention, the re-execution of the recognition process includes:

[0026] a) using the acquired images of the decoction pieces to be decocted to be re-input into the convolutional neural network recognition model for recognition;

[0027] b) Remove the previously confirmed names or identification marks from the returned identification results, and use the remaining names or identification marks as query conditions to continue to determine whether there is a data row for the name or identification mark in the database:

[0028] If it does not exist, re-obtain the image of the decoction piece to be decocted, use the newly acquired image of the decoction piece to be decocted to input into the convolutional neural network recognition model for recognition, and continue to perform the above step b). If it still does not exist, mark the decoction piece to be decocted before administration;

[0029] If it exists, then obtain the data of all data rows of the corresponding boiled and loose medicinal pieces to which the name or identification mark belongs.

[0030] As a further improvement of the present invention, the information on the prescription sheet is compared with the data of each data row corresponding to the boiled and dispersed decoction pieces, including:

[0031] The decoction system reads the Chinese medicine information on the prescription and writes different information in the Chinese medicine information into a comparison frame of a specific format, including at least specific prescription ingredients, names of each ingredient, usage and dosage information, and decoction information;

[0032] For the information in the comparison box, a comparison is performed from the data of the obtained boiled and loose medicinal pieces, including whether the recognition results contain all the specific prescription ingredients on the prescription and whether the name information of each ingredient is consistent, and whether the usage and dosage information on the prescription exceeds the standard usage and dosage recorded in the data.

[0033] As a further improvement of the present invention, the information on the prescription is compared with the data of each data row corresponding to the boiled and dispersed decoction pieces:

[0034] When a name discrepancy occurs, that is, all the boiled and loose pieces obtained from the database corresponding to the identification result are different from the specific prescription ingredients on the prescription and / or the obtained boiled and loose pieces name information is inconsistent with the name information of each ingredient on the prescription, the boiled and loose pieces to be decocted shall be marked before administration.

[0035] As a further improvement of the present invention, the information on the prescription is compared with the data of each data row corresponding to the boiled and dispersed decoction pieces:

[0036] When an overdose occurs, that is, the usage and dosage information on the prescription exceeds the usage and dosage information of the corresponding quality standard of the Chinese medicine decocted and dispersed pieces recorded in the database, the decocted and dispersed pieces to be decocted shall be marked before administration.

[0037] Compared with the prior art, the present invention has the following beneficial effects: by identifying the actual condition of the specific boiled and dispersed pieces during the decoction and comparing them with the information on the prescription, omissions or inconsistent or excessive medication can be avoided, thereby improving the accuracy of adjustment and ensuring the safety of patient medication. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the technical solution, the drawings required for use in the implementation will be briefly introduced below. Obviously, the drawings described below are only some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0039] Figure 1 It is a schematic diagram of a flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to enable a clear and complete understanding of the technical solution, the present invention is further described in conjunction with the embodiments and drawings. Obviously, the described embodiments are only some embodiments of the present invention, and all other embodiments obtained by technicians in the relevant field without making creative work are within the scope of protection of the present invention.

[0041] It should be understood that when used in this specification and the appended claims, the terms "include" and "comprises" indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0042] It should also be understood that the terms used in this specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the specification of the present invention and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0043] It should be further understood that the term "and / or" used in the present description and the appended claims refers to any and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0044] The embodiment of the present invention provides a method for identifying boiled and loose Chinese medicine slices based on artificial intelligence, which is aimed at identifying the Chinese medicine information on the prescription and dispensing the medicine according to the Chinese medicine information, identifying and checking the boiled and loose Chinese medicine slices that are about to enter the decoction after the medicine is dispensed, identifying the specific boiled and loose Chinese medicine slices on the medicine conveying line, and checking them with the information on the prescription, so as to avoid omissions or inconsistent or excessive medication, improve the accuracy of adjustment and ensure the safety of patient medication.

[0045] Specifically, a method for identifying Chinese herbal medicine pieces based on artificial intelligence, such as Figure 1 As shown, the following steps are included:

[0046] Acquire the image of the decoction pieces to be decocted; after reading the Chinese medicine information on the prescription, the decoction system controls the medicine conveying line to pick up the medicine, and collects the image of each decoction piece in the medicine conveying line;

[0047] Inputting the image of the boiled and scattered medicinal pieces into a convolutional neural network recognition model for recognition, and returning a recognition result, wherein the convolutional neural network recognition model is pre-trained;

[0048] Traversing the database, and obtaining data of each data row corresponding to the boiled and dispersed medicinal pieces according to the recognition result;

[0049] Compare the data on the prescription with the data in each row of the corresponding decocted herbal slices. If there is any name discrepancy and / or excess, mark it before the decocted herbal slices are administered and dispensed.

[0050] In this embodiment, the images of boiled and loosened medicinal pieces are recognized by a convolutional neural network recognition model. The images of boiled and loosened medicinal pieces may include multiple images of boiled and loosened medicinal pieces or one of the images of boiled and loosened medicinal pieces. If there are multiple images of boiled and loosened medicinal pieces, they need to be relatively scattered. For this, they can be set to be stacked separately during the medicine picking process, or each boiled and loosened medicinal piece needs to be weighed when picking the medicine, and each independent image of the boiled and loosened medicinal piece can be obtained during the weighing process, or the medicine can be picked up uniformly by vibrating or prying apart the stacked boiled and loosened medicinal pieces, etc., all of which can achieve the acquisition of relatively scattered or stacked images of each boiled and loosened medicinal piece. This embodiment does not impose any form of restriction on this.

[0051] The convolutional neural network recognition model is used to pre-process the image of the boiled and scattered medicinal pieces, such as graying and Gaussian filtering, and then the image is segmented, and the image features are identified. The recognition process adopts existing technical means. After the recognition, the recognition result is returned, and the recognition result can identify which boiled and scattered medicinal pieces are present, and obtain the name or identification mark of the identified boiled and scattered medicinal pieces (which can be the index or code calibrated for each boiled and scattered medicinal piece in the training recognition model, etc., which is unique).

[0052] In an optional embodiment, the convolutional neural network recognition model includes:

[0053] Collect standard sample images of boiled and loose medicinal pieces;

[0054] The characteristic information of the standard sample images is calibrated through the properties and identification methods of each boiled and dispersed herbal slice. The properties and identification methods of each boiled and dispersed herbal slice shall be based on the currently published group standards such as the guiding principles and technical requirements for the research on the quality standards of boiled and dispersed herbal slices of Chinese medicine. For example, the notice issued by the Office of the Guangdong Provincial Food and Drug Administration on the review and issuance of the "Guiding Principles and Technical Requirements for the Research on the Quality Standards of Boiled and Dispersed Herbal Slices of Chinese Medicine (Trial)" clearly stipulates that the text of the quality standards for boiled and dispersed herbal slices shall be written in the order of name, source, preparation method of boiled and dispersed herbal slices, properties, identification (including microscopic identification, thin layer identification or DNA barcode identification), inspection, extract, content determination, nature and flavor and meridians, functions and indications, usage and dosage, precautions, and storage.

[0055] For example, the [properties] of Chuanlianzi decoction pieces record: This product is in irregular small pieces or granules, with a particle size range of 0.8mm to 6.7mm. The surface is light yellow, golden yellow to brownish yellow, and the outer peel is brown and leathery. Black-brown broken seeds are occasionally seen. The smell is peculiar, and the taste is sour and bitter; Regarding [identification], it is recorded: The powder of this product is yellow-brown. The pericarp fibers are bundled, blunt at the end, 9 to 36μm in diameter, and the wall is very thick. The surrounding thin-walled cells contain calcium oxalate prismatic crystals to form crystal fibers. The pericarp stone cells are quasi-circular, irregularly long strips or long polygons, some with tumor-like protrusions or blunt short branches, with a diameter of 14 to 54μm and a length of about 150μm. The seed coat cells are bright yellow or orange-yellow, and under the epidermis is a row of quasi-square cells, with a diameter of about 44μm, very thick walls, and longitudinal microwave textures, connected to the pigment layer below. The surface of the epidermal cells is polygonal, with a dense granular texture. The cavity of the cells in the pigment layer of the seed coat is filled with reddish brown substances. The diameter of the crystal-containing cells in the seed coat is 13 to 27 μm, with varying wall thicknesses. Thick ones form stone cells, and the cavity is filled with light yellow, yellow-brown or reddish brown substances, and contains tiny calcium oxalate prismatic crystals with a diameter of about 5 μm. The diameter of the calcium oxalate cluster crystals is 5 to 27 μm.

[0056] In this embodiment, the characteristic information is formed by calibrating the standard sample image through the properties and identification method of the decoction pieces of Chuanlianzi, that is, the properties and identification method belong to the known mature identification method, and the characteristic calibration refers to the use of software to identify the characteristic position and / or mark the characteristic position for learning and training. For example, the stone cells of the pericarp are quasi-circular, irregularly long strips or long polygons, and some have tumor-like protrusions or blunt short branches, etc., and the characteristic positions of the relevant images are calibrated.

[0057] Further, the characteristic information of each boiled and scattered herbal slice is imported into a preset machine learning model for training to obtain a convolutional neural network prediction model. In some embodiments, the background is required to be a solid color background when collecting images, at least 100 different images are taken for each boiled and scattered herbal slice, the image size and image ratio are set, and the state or shooting angle height is arbitrarily changed for each image taken, so as to further improve the generalization ability and accuracy of the model. In this regard, this embodiment does not impose any restrictions on the principle and implementation method of model training. As long as the existing technology can be used to obtain a recognition model that can identify specific boiled and scattered herbal slices through training with a large number of sample images, it will be sufficient.

[0058] The convolutional neural network prediction model is evaluated based on the set model evaluation parameter index. If the evaluation passes, the convolutional neural network recognition model is obtained. It should be noted that this embodiment does not impose any restrictions on the model evaluation parameter index and the implementation means of model evaluation. The technicians can set them according to actual use requirements. As long as the model evaluation can be achieved through existing technical means, the convolutional neural network recognition model of this embodiment can be obtained.

[0059] Furthermore, the image of the boiled and scattered medicinal pieces is input into the convolutional neural network recognition model for recognition, including:

[0060] The image of the boiled and scattered medicinal pieces contains at least one pixel of boiled and scattered medicinal pieces;

[0061] The images of the boiled and scattered medicinal pieces are input into the trained convolutional neural network recognition model, and the recognition results of the names or identification marks of the boiled and scattered medicinal pieces are output.

[0062] In an optional embodiment, obtaining data of each data row corresponding to the boiled and dispersed medicinal pieces includes:

[0063] Obtain the name or identification mark of the corresponding boiled and dispersed medicinal piece from the recognition result, and use the name or identification mark as a query condition to determine whether there is a data row of the name or the identification mark in the database;

[0064] If it does not exist, the identification process is re-executed;

[0065] If it exists, then obtain the data of all data rows of the corresponding boiled and loose medicinal pieces to which the name or identification mark belongs.

[0066] In the above steps, re-performing the recognition process includes:

[0067] a) using the acquired images of the decoction pieces to be decocted to be re-input into the convolutional neural network recognition model for recognition;

[0068] b) Remove the previously confirmed names or identification marks from the returned identification results, and use the remaining names or identification marks as query conditions to continue to determine whether there is a data row for the name or identification mark in the database:

[0069] If it does not exist, re-obtain the image of the decoction piece to be decocted, use the newly acquired image of the decoction piece to be decocted to input into the convolutional neural network recognition model for recognition, and continue to perform the above step b). If it still does not exist, mark the decoction piece to be decocted before administration;

[0070] If it exists, then obtain the data of all data rows of the corresponding boiled and loose medicinal pieces to which the name or identification mark belongs.

[0071] In this embodiment, the database is the traceability information pre-connected to the traceability system of Chinese medicine decoction pieces so that the current database stores all the decoction pieces information that can currently be prescribed, and the decoction pieces information includes the information on the species, breeding, production, processing, sales, transportation, warehousing and quality standards of the decoction pieces corresponding to the decoction pieces;

[0072] According to each item of information of the boiled and dispersed medicinal slices, corresponding data items and identification marks for searching or indexing are formed.

[0073] Generally speaking, there are relevant regulations on Chinese herbal medicine pieces, which require full-process quality control. Therefore, when purchasing or warehousing corresponding Chinese herbal medicine pieces, there is a corresponding Chinese herbal medicine piece traceability system to record its traceability information, that is, when each batch of Chinese herbal medicine pieces are purchased in a hospital or institution, the whole process information of the pieces can be obtained, including breeding information, picking information, processing and sales information, warehousing information, etc. The database in this embodiment completes the docking of relevant information when the relevant pieces are put into storage, including the quality standard information corresponding to the pieces, which is the quality standard that the relevant drug inspection institute has determined that the review method is feasible and the standard content is publicized to the society, and the name, properties and other information of the corresponding pieces are recorded in the quality standard content. For the information of each kind of pieces in this database, corresponding data items and identification marks for searching or indexing are formed, and the corresponding data can be found in the database through the relevant name or identification mark.

[0074] Through the above steps, this embodiment obtains the name or identification mark of the corresponding boiled and loose medicinal pieces from the recognition result, and uses the name or identification mark as the query condition to obtain the data of the corresponding boiled and loose medicinal pieces. If it cannot be found, it means that the boiled and loose medicinal pieces on the medicine conveying line and the boiled and loose medicinal pieces data in the database of this hospital and this institution are abnormal, and there is a risk of medication. Corresponding markings are made and manual review is required, which greatly improves work efficiency while ensuring the safety of medication.

[0075] In an optional embodiment, the step of comparing the information on the prescription sheet with the data of each data row corresponding to the boiled and dispersed decoction piece comprises:

[0076] The decoction system reads the Chinese medicine information on the prescription and writes different information in the Chinese medicine information into a comparison frame of a specific format, including at least specific prescription ingredients, names of each ingredient, usage and dosage information, and decoction information;

[0077] For the information in the comparison box, a comparison is performed from the data of the obtained boiled and loose medicinal pieces, including whether the recognition results contain all the specific prescription ingredients on the prescription and whether the name information of each ingredient is consistent, and whether the usage and dosage information on the prescription exceeds the standard usage and dosage recorded in the data.

[0078] For example, the information of traditional Chinese medicine obtained in the current prescription forms an ingredient comparison box, which includes A boiled and loosened slices, B boiled and loosened slices, and C boiled and loosened slices, and a dosage comparison box is formed after each boiled and loosened slice. This is used to obtain the data corresponding to the recognition result in the database and fill it into the comparison box. If there are abnormalities such as missing ingredients, inconsistent ingredients, and excessive dosage, there is a risk of medication, and corresponding marking is made, which requires manual review.

[0079] For better understanding, in an optional embodiment, the information on the prescription is compared with the data of each data row corresponding to the boiled and dispersed decoction pieces:

[0080] For the comparison frame formed by the Chinese medicine information on the prescription, the data of the corresponding boiled and loose pieces of medicine are introduced for comparison;

[0081] When a name discrepancy occurs, that is, all the boiled and loose pieces obtained from the database corresponding to the identification result are different from the specific prescription ingredients on the prescription and / or the obtained boiled and loose pieces name information is inconsistent with the name information of each ingredient on the prescription, the boiled and loose pieces to be decocted shall be marked before administration.

[0082] For better understanding, in an optional embodiment, the information on the prescription is compared with the data of each data row corresponding to the boiled and dispersed decoction pieces:

[0083] According to the usage and dosage information of the corresponding boiled and scattered pieces of medicine from the recognition results, the usage and dosage of the corresponding boiled and scattered pieces of medicine from the prescription information are introduced for comparison;

[0084] When an overdose occurs, that is, the usage and dosage information on the prescription exceeds the usage and dosage information of the corresponding quality standard of the Chinese medicine decocted and dispersed pieces recorded in the database, the decocted and dispersed pieces to be decocted shall be marked before administration.

[0085] The present invention can identify and verify the verification link before decoction after the medicine is picked, and can avoid omissions or inconsistent or excessive medication by identifying the actual conditions of the specific boiled and dispersed pieces during the decoction and verifying them with the information on the prescription, thereby improving the accuracy of adjustment and ensuring the safety of patients' medication.

[0086] The above disclosure is only one or more preferred embodiments of the present invention, which are used to help understand the inventive concept of the technical solution, and is not intended to limit the present invention in other forms. Technical personnel in the relevant field may make other equivalent or customary replacement schemes based on the features defined by the present invention, which still fall within the scope of the present invention.

Claims

1. A method for identifying Chinese medicine pieces based on artificial intelligence, characterized in that: The following steps are involved: Acquire the image of the decoction pieces to be decocted; Inputting the image of the boiled and scattered medicinal pieces into a convolutional neural network recognition model for recognition, and returning a recognition result, wherein the convolutional neural network recognition model is pre-trained; Traversing the database, and obtaining data of each data row corresponding to the boiled and dispersed medicinal pieces according to the recognition result; Compare the data on the prescription with the data in each row of the corresponding decocted herbal slices. If there is any name discrepancy and / or excess, mark it before the decocted herbal slices are administered and dispensed.

2. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 1, characterized in that: The method of obtaining the image of the decoction pieces to be decocted comprises: After reading the Chinese medicine information on the prescription, the decoction system controls the medicine conveyor line to pick up the medicine and collects the images of each boiled and loose medicine piece on the medicine conveyor line.

3. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 1, characterized in that: The convolutional neural network recognition model includes: Collect standard sample images of boiled and loose medicinal pieces; The characteristic information is formed by calibrating the standard sample image through the properties and identification methods of each boiled and loose medicinal piece; Import the characteristic information of each boiled and dispersed herbal medicine piece into a preset machine learning model for training to obtain a convolutional neural network prediction model; The convolutional neural network prediction model is evaluated based on the set model evaluation parameter indicators. If the evaluation passes, a convolutional neural network recognition model is obtained.

4. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 3, characterized in that: The step of inputting the boiled and scattered medicinal pieces image into the convolutional neural network recognition model for recognition comprises: The image of the boiled and scattered medicinal pieces contains at least one pixel of boiled and scattered medicinal pieces; The images of the boiled and scattered medicinal pieces are input into the trained convolutional neural network recognition model, and the recognition results of the names or identification marks of the boiled and scattered medicinal pieces are output.

5. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 1, characterized in that: The data of each data row corresponding to the boiled and dispersed decoction pieces is obtained as follows: Obtain the name or identification mark of the corresponding boiled and dispersed medicinal piece from the recognition result, and use the name or identification mark as a query condition to determine whether there is a data row of the name or the identification mark in the database; If it does not exist, the identification process is re-executed; If it exists, then obtain the data of all data rows of the corresponding boiled and loose medicinal pieces to which the name or identification mark belongs.

6. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 5, characterized in that: The database is pre-connected with the traceability information of the Chinese medicine boiled and scattered pieces traceability system so that the current database stores all the boiled and scattered pieces information that can currently be prescribed, and the boiled and scattered pieces information includes the information on the species and breeding, production and processing, sales and transportation, warehousing and the corresponding quality standards of the boiled and scattered pieces of Chinese medicine; According to each item of information of the boiled and dispersed medicinal slices, corresponding data items and identification marks for searching or indexing are formed.

7. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 5, characterized in that: The re-execution recognition process includes: a) using the acquired images of the decoction pieces to be decocted to be re-input into the convolutional neural network recognition model for recognition; b) Remove the previously confirmed names or identification marks from the returned identification results, and use the remaining names or identification marks as query conditions to continue to determine whether there is a data row for the name or identification mark in the database: If it does not exist, re-obtain the image of the decoction piece to be decocted, use the newly acquired image of the decoction piece to be decocted to input into the convolutional neural network recognition model for recognition, and continue to perform the above step b). If it still does not exist, mark the decoction piece to be decocted before administration; If it exists, then obtain the data of all data rows of the corresponding boiled and loose medicinal pieces to which the name or identification mark belongs.

8. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 1, characterized in that: The information on the prescription sheet according to the data comparison of each data row corresponding to the boiled and dispersed decoction pieces includes: The decoction system reads the Chinese medicine information on the prescription and writes different information in the Chinese medicine information into a comparison frame of a specific format, including at least specific prescription ingredients, names of each ingredient, usage and dosage information, and decoction information; For the information in the comparison box, a comparison is performed from the data of the obtained boiled and loose medicinal pieces, including whether the recognition results contain all the specific prescription ingredients on the prescription and whether the name information of each ingredient is consistent, and whether the usage and dosage information on the prescription exceeds the standard usage and dosage recorded in the data.

9. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 8, characterized in that: Compare the information on the prescription with the data in each row of the corresponding boiled and loose pieces: When a name discrepancy occurs, that is, all the boiled and loose pieces obtained from the database corresponding to the identification result are different from the specific prescription ingredients on the prescription and / or the obtained boiled and loose pieces name information is inconsistent with the name information of each ingredient on the prescription, the boiled and loose pieces to be decocted shall be marked before administration.

10. The method for identifying Chinese herbal medicine pieces based on artificial intelligence according to claim 8, characterized in that: Compare the information on the prescription with the data in each row of the corresponding boiled and loose pieces: When an overdose occurs, that is, the usage and dosage information on the prescription exceeds the usage and dosage information of the corresponding quality standard of the Chinese medicine decoction pieces recorded in the database, the decoction pieces to be decocted shall be marked before administration.

Images