Method, system, equipment and medium for connecting species identification experts based on AI recognition

Through the combination of AI recognition model and expert database, the automatic allocation of species identification is achieved, solving the problem of insufficient accuracy of species identification in the prior art, and improving the recognition efficiency and quality.

CN117456278BActive Publication Date: 2024-11-26SOUTH CHINA BOTANICAL GARDEN CHINESE ACADEMY OF SCI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311623097.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-29
Publication Date
2024-11-26
Estimated Expiration
2043-11-29

AI Technical Summary

Technical Problem

The existing AI recognition model based on deep learning is difficult to achieve the accuracy of species and below in plant species recognition, especially the recognition effect of wild plants is not ideal, resulting in the reliance on experts in species identification and lack of automated allocation mechanisms.

Method used

The AI ​​identification model obtains species identification results, combines target position data, and screens out matching expert lists from the expert database to realize automatic allocation of species identification and improve identification efficiency and quality.

Benefits of technology

Automatic allocation of species identification is realized, the efficiency and quality of species identification is improved, and the accuracy and professionalism of identification results are ensured.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117456278B_ABST
    Figure CN117456278B_ABST
Patent Text Reader

Abstract

The present invention discloses a species identification expert docking method, system, device and medium based on AI recognition. The key points of its technical solution are: obtaining target data of a target species, wherein the target data includes: target species data and target location data; inputting the target species data into a trained AI recognition model to obtain a recognition result, and obtaining a corresponding target classification level according to the recognition result; screening out a corresponding target expert list from an expert database according to the target classification level or target location data; sending the target species data to experts in the target expert list; receiving identification feedback from experts in the target expert list, determining the optimal identification feedback according to the identification feedback of each expert in the target expert list, realizing automatic allocation of species identification tasks, and being able to greatly improve the efficiency and quality of species identification.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of AI recognition technology, and specifically relates to a species identification expert docking method, system, equipment and medium based on AI recognition. Background Art

[0002] At present, in the field of biological species identification, AI recognition models constructed using a large amount of biodiversity data (such as biological photos and biological sounds) are also widely used in the classification and auxiliary identification of various biological species. After deep learning, the AI ​​recognition model can realize functions such as identifying flowers and plants by taking photos, and identifying animals by hearing sounds. Users input audio data, image data or video data into the AI ​​model, and the AI ​​model can quickly identify biological species and return at least one recognition result.

[0003] The existing AI recognition models based on deep learning still have difficulty in ensuring the accuracy of species identification. For example, for large-scale plant species identification, in practical applications, AI recognition models can usually only achieve good results in the definition of the genus level, while the identification of species and below has great uncertainty, especially for wild plants, the identification effect of species is often not ideal. Therefore, the accurate identification of species currently still relies mainly on various experts, but how to automatically assign identification experts for species identification to improve identification efficiency remains to be studied. Summary of the invention

[0004] The purpose of the present invention is to provide a species identification expert docking method, system, equipment and medium based on AI recognition, so as to realize the automatic allocation of species identification tasks and greatly improve the efficiency and quality of species identification.

[0005] The first aspect of the present invention discloses a species identification expert docking method based on AI recognition, comprising:

[0006] Acquire target data of a target species, wherein the target data includes: target species data and target location data;

[0007] Input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result;

[0008] Filtering a corresponding target expert list from an expert database according to the target classification level or target location data;

[0009] Sending the target species data to experts in the target expert list;

[0010] Receive the evaluation feedback from the experts in the target expert list, and determine the optimal evaluation feedback according to the evaluation feedback from each expert in the target expert list.

[0011] Optionally, the expert database is a database storing expert information on various species identification aspects, wherein the expert information includes: expert ID, identification field to which the expert belongs, taxonomic group to which the expert belongs, and geographical area of ​​concern to which the expert focuses; the corresponding target expert list is selected from the expert database according to the target taxonomic level or target location data, including:

[0012] Selecting experts whose taxonomic groups include target taxonomic elements from the expert database to obtain a first expert list;

[0013] Selecting experts whose taxonomic group includes a first identification taxonomic class from the expert database to obtain a second expert list, wherein the first identification taxonomic class is the taxonomic class to which the target taxonomic class belongs;

[0014] Selecting experts whose geographical area of ​​interest includes the target location data and whose taxonomic group includes the second identification taxonomic category from the expert database to obtain a third expert list, wherein the second identification taxonomic category is the taxonomic category to which the first identification taxonomic category belongs;

[0015] A target expert list is selected from the first expert list, the second expert list and the third expert list according to the first preset instruction.

[0016] Optionally, selecting a target expert list from the first expert list, the second expert list, and the third expert list according to the first preset instruction includes:

[0017] Determine whether the number in the first expert list reaches a preset recommended number;

[0018] If yes, selecting a preset recommended number of experts from the first expert list as a target expert list;

[0019] If not, determine whether the sum of the number of experts in the second expert list reaches the preset recommended number. If so, select all the experts in the first expert list and the remaining experts in the second expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, determine whether the number of experts in the second expert list and the third expert list reaches the preset recommended number. If so, select all the experts in the second expert list and the remaining experts in the third expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, use all the experts in the second expert list and the third expert list as the target expert list.

[0020] Optionally, after selecting a corresponding target expert list from an expert database according to the target classification level or target location data, the following steps may be performed:

[0021] Experts in the target expert list are sorted according to preset priorities, and a preset number of target experts are selected from the front to the back according to the sorting in the target expert list, wherein the preset priorities include: probability of expert feedback, expert professional title level, or probability of users giving favorable comments on expert historical appraisals.

[0022] Optionally, determining the optimal evaluation feedback according to the evaluation feedback of the experts in the target expert list includes:

[0023] Counting the number of identical identification feedbacks, and taking the identification feedback with the largest number as the first identification feedback;

[0024] Comparing the relationship between each of the identification feedbacks and the corresponding expert's taxonomic group, and taking the identification feedback with the closest relationship to the corresponding expert's taxonomic group as the second identification feedback;

[0025] Compare the preset expert weights corresponding to the respective appraisal feedbacks, and take the appraisal feedback with the highest preset expert weight as the third appraisal feedback;

[0026] The optimal identification feedback is selected from the first identification feedback, the second identification feedback and the third identification feedback according to the second preset instruction.

[0027] Optionally, sending the target species data to experts in a target expert list includes:

[0028] Counting the number of times a message is sent to each expert in the target expert list within a preset time;

[0029] The target species data is sent to the experts whose sending times do not exceed a first preset threshold during the experts' free time.

[0030] Optionally, obtaining a corresponding target classification level according to the recognition result includes:

[0031] In the case where the recognition result is a classification element of species or below, a classification element to be selected is selected from the recognition result according to the confidence, user instruction or the classification element of the recognition result, and the classification element to which the classification element to be selected belongs is used as the target classification element;

[0032] When the recognition result is a genus or above classification element, a candidate classification element is selected from the recognition result according to the confidence, user instructions or the classification element of the recognition result, and the candidate classification element or the classification element to which it belongs is used as the target classification element.

[0033] The second aspect of the present invention discloses a species identification expert docking system based on AI recognition, comprising:

[0034] A data acquisition module, used to acquire target data of a target species, wherein the target data includes: target species data and target location data;

[0035] A species recognition module is used to input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result;

[0036] An expert screening module, used to screen out a corresponding target expert list from an expert database according to the target classification level or target location data;

[0037] A data sending module, used for sending the target species data to experts in the target expert list;

[0038] The feedback receiving and determining module is used to receive the appraisal feedback from the experts in the target expert list, and determine the optimal appraisal feedback according to the appraisal feedback from each expert in the target expert list.

[0039] A third aspect of the present invention discloses a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0040] A fourth aspect of the present invention discloses a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0041] The technical solution provided by the present invention has the following advantages and effects: first, the recognition result is identified by the AI ​​recognition model, and the target classification level is determined by the recognition result. Then, experts with high matching degree are screened from the expert database according to the target classification level or target position data to form an expert list, so as to select the corresponding experts to identify the target species, realize the automatic allocation of species identification tasks, and greatly improve the efficiency and quality of species identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of a method for connecting experts to species identification based on AI recognition disclosed in an embodiment of the present invention;

[0043] Figure 2 It is a structural block diagram of the species identification expert docking system based on AI recognition disclosed in an embodiment of the present invention;

[0044] Figure 3 It is a diagram of the internal structure of a computer device disclosed in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] In order to facilitate the understanding of the present invention, specific embodiments of the present invention will be described in more detail below with reference to the accompanying drawings.

[0046] Unless otherwise specified or defined, the "first, second..." used in this article is merely used to distinguish names and does not represent a specific quantity or order.

[0047] Unless specifically stated or defined otherwise, the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0048] It should be noted that, in this document, “fixed to” or “connected to” may mean directly fixing or connecting to an element, or indirectly fixing or connecting to an element.

[0049] like Figure 1 As shown, the species identification expert docking method based on AI recognition includes:

[0050] Step 1: Acquire target data of a target species, wherein the target data includes target species data and target location data.

[0051] In this embodiment, the target species data includes: any one or more of audio data, image data and video data containing the target species, and the target species may be plants, animals or fungi; the target species data may be currently recorded or photographed by the user, or may be pre-stored, and the currently recorded / photographed or pre-stored target species data may be directly uploaded by the user to the server, or may be collected by the user using applications pre-installed on various terminals (mobile phones, iPads), and the target location data is the geographic data of the location where the user discovers the target species, and the target location data may be directly uploaded by the user to the server, or may be collected by the user using applications pre-installed on various terminals, and the target location data may be represented by latitude and longitude coordinates, administrative divisions, geographic grid codes or address codes.

[0052] Step 2: Input the target species data into the trained AI recognition model to obtain a recognition result, and obtain the corresponding target classification level according to the recognition result.

[0053] In this embodiment, the AI ​​recognition model is a neural network-based model that can be deployed in the cloud or directly in the terminal application. The AI ​​recognition model includes: an audio recognition model and an image recognition model; after inputting the target species data, first determine the type of the target species data. If it is audio data, the target species data is input into the trained audio recognition model, and the trained audio recognition model recognizes the audio data to obtain the corresponding recognition result. If it is image data, the target species data is input into the trained image recognition model, and the trained image recognition model recognizes the image data to obtain the corresponding recognition result. If it is video data, the video frame in the video data is intercepted as the image data containing the target species, and the image data is input into the trained image recognition model to obtain the corresponding recognition result.

[0054] In this embodiment, the audio recognition model can adopt the general audio recognition model in the prior art, and the image recognition model can adopt the general image recognition model in the prior art. The general audio recognition model is trained to obtain a trained audio recognition model, and the general image recognition model is trained to obtain a trained image recognition model. Specifically, audio samples of each species are obtained as an audio training set, and each audio sample is marked with corresponding species-level, genus-level, family-level and / or order-level classification elements. The audio training set is used to train the general audio model until the general audio model converges or reaches a preset number of training rounds to obtain a trained audio recognition model. Image samples of each species are obtained as an image training set, and each image sample is marked with corresponding species-level, genus-level, family-level and / or order-level classification elements. The image training set is used to train the general image model until the general image model converges or reaches a preset number of training rounds to obtain a trained image recognition model.

[0055] Step 3: Filter out a corresponding target expert list from the expert database according to the target classification level or target location data.

[0056] In this embodiment, experts related to the target taxonomic category can be screened out from the expert database according to actual conditions to form a target expert list, or experts related to the target location data and the target taxonomic category can be screened out from the expert database to form a target expert list, so as to facilitate screening out experts related to the target species from the expert database, and then sending the target species data to the corresponding experts for professional identification.

[0057] Step 4: Send the target species data to the experts in the target expert list so that the experts in the target expert list can identify the target species. In this embodiment, the target taxonomic level can also be sent to the experts in the target expert list for reference.

[0058] Step 5: Receive the appraisal feedback from the experts in the target expert list, and determine the optimal appraisal feedback according to the appraisal feedback from each expert in the target expert list.

[0059] In this embodiment, the identification feedback is the expert's identification opinion on the target species data, which may be a specific species identification or may be impossible to identify. After the target species data is sent to the experts in the target expert list, some experts may not give identification feedback in time. As long as the experts who give the identification feedback meet the preset conditions, the optimal identification feedback can be determined by combining all the identification feedback given. In this embodiment, the recognition result is first identified by the AI ​​recognition model, and the target classification level is determined by the recognition result. Then, according to the target classification level or target location data, experts with high matching degree are screened from the expert library to form an expert list, so as to select the corresponding experts to identify the target species and realize the automatic allocation of species identification tasks, which can greatly improve the efficiency and quality of species identification.

[0060] Furthermore, the expert database is a database storing expert information on various species identification aspects, wherein the expert information includes: expert ID, identification field to which the expert belongs, taxonomic group to which the expert belongs, and geographical area of ​​concern to which the expert focuses; the corresponding target expert list is selected from the expert database according to the target taxonomic level or target location data, including:

[0061] Selecting experts whose taxonomic groups include target taxonomic elements from the expert database to obtain a first expert list;

[0062] Selecting experts whose taxonomic group includes a first identification taxonomic class from the expert database to obtain a second expert list, wherein the first identification taxonomic class is the taxonomic class to which the target taxonomic class belongs;

[0063] Selecting experts whose geographical area of ​​interest includes the target location data and whose taxonomic group includes the second identification taxonomic category from the expert database to obtain a third expert list, wherein the second identification taxonomic category is the taxonomic category to which the first identification taxonomic category belongs;

[0064] A target expert list is selected from the first expert list, the second expert list and the third expert list according to the first preset instruction.

[0065] In this embodiment, the expert information also includes: expert title level, expert age, expert feedback probability, user favorable evaluation probability of the expert's historical appraisal, etc. In other embodiments, the expert information can also include other information that reflects the expert's energy, ability and enthusiasm for participating in the appraisal, just like the expert feedback probability; the taxonomic group to which the expert belongs should cover the corresponding high-order taxonomic meta-information. For example, if an expert is good at identifying Rhododendron plants, then the taxonomic group information to which the expert belongs should include not only the Rhododendron genus, but also the high-order taxonomic meta-information such as the Ericaceae, Ericales, and Plant Kingdom to which the Rhododendron genus belongs. The taxonomic groups of the experts in the expert database may be the same or different. Then, according to the target taxonomic element, experts whose taxonomic groups include the target taxonomic element are screened out from the expert database to form a first expert list. For example, if the target taxonomic element is the genus Dalbergia of the Leguminosae family, experts whose taxonomic group is the genus Dalbergia of the Leguminosae family are screened out from the expert database to form a first expert list, so as to obtain more accurate and professional experts. The taxonomic element to which the genus Dalbergia of the Leguminosae family belongs is the Leguminosae family, so the first identification taxonomic element is the Leguminosae family, and experts whose taxonomic group is the Leguminosae family are screened out from the expert database to form a second expert list. Experts whose taxonomic group is the Leguminosae family include experts of the genus Dalbergia and other genera of the Leguminosae family, so as to obtain more experts. The geographical area of ​​interest may be represented by latitude and longitude coordinates, administrative divisions, geographic grid codes or address codes. When selecting experts from the expert database according to the target location data, first determine whether the representation of the target location data is the same as that of the geographical area of ​​interest. If different, convert the target location data into a taxonomic element that is consistent with the geographical area of ​​interest. The same representation method is used for the geographical area of ​​interest, and then experts whose geographical area of ​​interest contains the target location data and whose taxonomic group contains the second identification taxonomic level are screened out from the expert database to obtain a third expert list. If they are the same, experts whose geographical area of ​​interest contains the target location data and whose taxonomic group contains the second identification taxonomic level are directly screened out from the expert database to obtain a third expert list. For example, the target location data is Guangzhou, the AI ​​recognition result is white pheasant, the target taxonomic level is the genus Pheasant, and the first identification taxonomic level is the family Phasianidae. If there is no expert of the genus Pheasant and its first identification taxonomic level in the expert database, it is necessary to screen out experts of the second identification taxonomic level from the expert database according to the target location. The second identification taxonomic level can be Galliformes or Aves, but cannot be an expert in a non-bird field. For example, bird data cannot be sent to an expert in the plant field in Guangzhou. Therefore, the highest level of the second identification taxonomic level that can be set needs to be limited by the system according to the characteristics of the expert group of each biological taxonomic group.

[0066] Further, selecting a target expert list from the first expert list, the second expert list, and the third expert list according to the first preset instruction includes:

[0067] Determine whether the number in the first expert list reaches a preset recommended number;

[0068] If yes, selecting a preset recommended number of experts from the first expert list as a target expert list;

[0069] If not, determine whether the sum of the number of experts in the second expert list reaches the preset recommended number. If so, select all the experts in the first expert list and the remaining experts in the second expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, determine whether the number of experts in the second expert list and the third expert list reaches the preset recommended number. If so, select all the experts in the second expert list and the remaining experts in the third expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, use all the experts in the second expert list and the third expert list as the target expert list.

[0070] In actual applications, the situation may occur in the expert database that the number of experts whose taxonomic group includes the target taxonomic category does not reach the preset recommended number. For example, the target taxonomic category is Dalbergia of the Leguminosae family, but the number of experts whose taxonomic group is Dalbergia of the Leguminosae family in the expert database does not reach the preset recommended number. In this case, it is necessary to screen experts whose taxonomic group includes the first identification taxonomic category from the expert database, that is, to screen experts in the Leguminosae taxonomic group. If the number of experts in the Leguminosae taxonomic group in the expert database reaches the preset recommended number, since the second expert list includes all the experts in the first expert list, all the experts in the first expert list and the remaining experts in the second expert list are selected to make the number of experts reach the preset recommended number, and the selected experts are used as the target expert list. If the sum of the number of experts in the Leguminosae taxonomic group and the number of experts whose geographical area of ​​interest includes the target location data in the expert database reaches the preset recommended number, all the experts in the second expert list and the remaining experts in the third expert list are selected to make the number of experts reach the preset recommended number, and the selected experts are used as the target expert list. In this embodiment, experts in the first expert list are preferentially selected as experts in the target expert list, which can provide more accurate and professional experts.

[0071] Further, after selecting a corresponding target expert list from an expert database according to the target classification level or target location data, the method further comprises:

[0072] The experts in the target expert list are sorted according to preset priorities, wherein the preset priorities include: expert feedback probability or expert professional title level.

[0073] In this embodiment, after obtaining the target expert list, the experts in the target expert list can be sorted in order of expert feedback probability from high to low, or the experts in the target expert list can be sorted in order of expert professional title level from high to low, so that experts with high expert feedback probability or high expert professional title level are ranked first.

[0074] Further, determining the optimal evaluation feedback according to the evaluation feedback of the experts in the target expert list includes:

[0075] Counting the number of identical identification feedbacks, and taking the identification feedback with the largest number as the first identification feedback;

[0076] Comparing the relationship between each of the identification feedbacks and the corresponding expert's taxonomic group, and taking the identification feedback with the closest relationship to the corresponding expert's taxonomic group as the second identification feedback;

[0077] Compare the preset expert weights corresponding to the respective appraisal feedbacks, and take the appraisal feedback with the highest preset expert weight as the third appraisal feedback;

[0078] The optimal identification feedback is selected from the first identification feedback, the second identification feedback and the third identification feedback according to the second preset instruction.

[0079] In this embodiment, after obtaining the identification feedback of multiple experts, such as the first identification feedback is Dalbergia clitoridis, the second identification feedback is Dalbergia clitoridis, the third identification feedback is Dalbergia angustifolia, the fourth identification feedback is Dalbergia clitoridis and the fifth identification feedback is unable to identify, then the Dalbergia clitoridis with the largest number is taken as the first identification feedback; when the target expert list is the second expert list or the third expert list, the identification fields of the experts in the target expert list may be different, such as the identification fields of the experts corresponding to the first identification feedback and the second identification feedback are both Dalbergia genus of the Leguminosae family, the identification field of the expert corresponding to the third identification feedback is Dalbergia genus of the Leguminosae family, the identification field of the expert corresponding to the fourth identification feedback is Rhododendron genus of the Leguminosae family, and the identification field of the expert corresponding to the fifth identification feedback is Rosaceae, then the identification fields of the experts corresponding to the first identification feedback and the second identification feedback are closest to the identification feedback, and Dalbergia clitoridis is taken as the second identification feedback; specifically, the expert weights of each expert in the expert database can be preset, such as the preset expert weights of the expert corresponding to the first identification feedback The expert weight is 80%, the preset expert weight of the expert corresponding to the second appraisal feedback is 60%, the preset expert weight of the expert corresponding to the third appraisal feedback is 50%, the preset expert weight of the expert corresponding to the fourth appraisal feedback is 50%, and the preset expert weight of the expert corresponding to the fifth appraisal feedback is 40%, then the first appraisal feedback, that is, Dalbergia obliquus, is used as the third appraisal feedback; the second preset instruction may be that when the target expert list is the first expert list, the first appraisal feedback can be used as the optimal appraisal feedback, when the target expert list is the second expert list, the second appraisal feedback can be used as the optimal appraisal feedback, when the target expert list is the third expert list, the third appraisal feedback can be used as the optimal appraisal feedback, when the target expert list is the first expert list and the number of experts in the target expert list is two, the third appraisal feedback can be used as the optimal appraisal feedback, that is, the optimal appraisal feedback can be selected from the first appraisal feedback, the second appraisal feedback and the third appraisal feedback according to the actual situation.

[0080] Further, sending the target species data to experts in the target expert list includes:

[0081] Counting the number of times a message is sent to each expert in the target expert list within a preset time;

[0082] The target species data is sent to the experts whose sending times do not exceed a first preset threshold during the experts' free time.

[0083] In this embodiment, in order to prevent excessive disturbance to the experts, it is necessary to limit the number of times messages are sent to the experts within a preset time. The preset time can be daily, and the first preset threshold can be set to three times. After obtaining the target expert list, the number of times messages have been sent to each expert in the target expert list within the preset time is counted, and messages are sent to the experts during the experts' free time, such as 11 a.m. or 4 p.m. to 9 p.m., etc. If there are experts A, B and C in the target expert list, the number of times expert A sends messages is 2 times, the number of times expert B sends messages is 3 times, and the number of times expert C sends messages is 0 times, then the number of times experts A and C send messages does not exceed the first preset threshold, and the target species data is sent to experts A and C, so that experts A and C identify the target species data. In other embodiments, the preset time can be every three days, every week, etc., and the first preset threshold can be increased or decreased according to actual conditions. The ways to send messages include: SMS, WeChat public account messages, mobile APP notifications, Web system messages, etc. The messages sent include the summary of the target species data and the access address. All the target species data of the target species can be obtained through the access address, and the target taxonomic level can also be sent to the expert. In addition, if there are multiple target species, for target species that have not received identification feedback for a long time, the connection with new experts will be postponed, and the recent target species data will be sent to the experts first.

[0084] Furthermore, obtaining the corresponding target classification level according to the recognition result includes:

[0085] In the case where the recognition result is a classification element of species or below, a classification element to be selected is selected from the recognition result according to the confidence, user instruction or the classification element of the recognition result, and the classification element to which the classification element to be selected belongs is used as the target classification element;

[0086] When the recognition result is a genus or above classification element, a candidate classification element is selected from the recognition result according to the confidence, user instructions or the classification element of the recognition result, and the candidate classification element or the classification element to which it belongs is used as the target classification element.

[0087] In this embodiment, there are usually multiple recognition results. When the recognition result is a species or lower level, the AI ​​recognition model will output the corresponding confidence level while outputting the recognition result. The recognition result with a confidence level higher than the second preset threshold can be used as a candidate classification level, or the recognition result with the highest confidence level can be used as a candidate classification level, or according to user instructions, the user instruction is the selection made by the user through the application, that is, the user selects the candidate classification level from the recognition result, or the classification level to which the recognition result belongs is recorded as a statistical classification level, the number of each statistical classification level is counted, and the recognition result corresponding to the statistical classification level with the largest number is used as the candidate classification level, and then the classification level to which the candidate classification level belongs is used as the target classification level; when the recognition result is a genus or higher level, after the candidate classification level is selected from the recognition result according to the confidence level, user instruction or the classification level of the recognition result, the candidate classification level can be directly used as the target classification level, or the classification level to which the candidate classification level belongs can be used as the target classification level.

[0088] The AI-based species identification expert docking method disclosed in the embodiment of the present invention first identifies the recognition result through the AI ​​recognition model, determines the target classification level through the recognition result, and then selects experts with high matching degree from the expert database according to the target classification level or target position data to form an expert list, so as to facilitate the selection of corresponding experts to identify the target species, realize the automatic allocation of species identification tasks, and can greatly improve the efficiency and quality of species identification.

[0089] like Figure 2 As shown, the present invention also provides a species identification expert docking system based on AI recognition, comprising:

[0090] The data acquisition module 10 is used to acquire target data of a target species, wherein the target data includes: target species data and target location data;

[0091] The species recognition module 20 is used to input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result;

[0092] An expert screening module 30, for screening out a corresponding target expert list from an expert database according to the target classification level or target location data;

[0093] A data sending module 40, used to send the target species data to the experts in the target expert list;

[0094] The feedback receiving and determining module 50 is used to receive the evaluation feedback from the experts in the target expert list, and determine the optimal evaluation feedback according to the evaluation feedback from each expert in the target expert list.

[0095] For the specific composition of the species identification expert docking system based on AI recognition, please refer to the composition of the species identification expert docking method based on AI recognition above, which will not be repeated here. The various modules of the above-mentioned species identification expert docking system based on AI recognition can be implemented in whole or in part through software, hardware and a combination thereof. The above-mentioned 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.

[0096] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a species identification expert docking method based on AI recognition is implemented.

[0097] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0098] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0099] Acquire target data of a target species, wherein the target data includes: target species data and target location data;

[0100] Input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result;

[0101] Filtering a corresponding target expert list from an expert database according to the target classification level or target location data;

[0102] Sending the target species data to experts in the target expert list;

[0103] Receive the evaluation feedback from the experts in the target expert list, and determine the optimal evaluation feedback according to the evaluation feedback from each expert in the target expert list.

[0104] In one embodiment, the expert database is a database storing expert information on various species identification aspects, wherein the expert information includes: expert ID, identification field to which the expert belongs, taxonomic group to which the expert belongs, and geographical area of ​​concern to which the expert focuses; the corresponding target expert list is selected from the expert database according to the target taxonomic level or target location data, including:

[0105] Selecting experts whose taxonomic groups include target taxonomic elements from the expert database to obtain a first expert list;

[0106] Selecting experts whose taxonomic group includes a first identification taxonomic class from the expert database to obtain a second expert list, wherein the first identification taxonomic class is the taxonomic class to which the target taxonomic class belongs;

[0107] Selecting experts whose geographical area of ​​interest includes the target location data and whose taxonomic group includes the second identification taxonomic category from the expert database to obtain a third expert list, wherein the second identification taxonomic category is the taxonomic category to which the first identification taxonomic category belongs;

[0108] A target expert list is selected from the first expert list, the second expert list and the third expert list according to the first preset instruction.

[0109] In one embodiment, selecting a target expert list from the first expert list, the second expert list, and the third expert list according to the first preset instruction includes:

[0110] Determine whether the number in the first expert list reaches a preset recommended number;

[0111] If yes, selecting a preset recommended number of experts from the first expert list as a target expert list;

[0112] If not, determine whether the sum of the number of experts in the second expert list reaches the preset recommended number. If so, select all the experts in the first expert list and the remaining experts in the second expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, determine whether the number of experts in the second expert list and the third expert list reaches the preset recommended number. If so, select all the experts in the second expert list and the remaining experts in the third expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, use all the experts in the second expert list and the third expert list as the target expert list.

[0113] In one embodiment, after selecting a corresponding target expert list from an expert database according to the target classification level or target location data, the following steps are performed:

[0114] Experts in the target expert list are sorted according to preset priorities, and a preset number of target experts are selected from the front to the back according to the sorting in the target expert list, wherein the preset priorities include: probability of expert feedback, expert professional title level, or probability of users giving favorable comments on expert historical appraisals.

[0115] In one embodiment, determining the optimal evaluation feedback according to the evaluation feedback of the experts in the target expert list includes:

[0116] Counting the number of identical identification feedbacks, and taking the identification feedback with the largest number as the first identification feedback;

[0117] Comparing the relationship between each of the identification feedbacks and the corresponding expert's taxonomic group, and taking the identification feedback with the closest relationship to the corresponding expert's taxonomic group as the second identification feedback;

[0118] Compare the preset expert weights corresponding to the respective appraisal feedbacks, and take the appraisal feedback with the highest preset expert weight as the third appraisal feedback;

[0119] The optimal identification feedback is selected from the first identification feedback, the second identification feedback and the third identification feedback according to the second preset instruction.

[0120] In one embodiment, sending the target species data to an expert in a target expert list includes:

[0121] Counting the number of times a message is sent to each expert in the target expert list within a preset time;

[0122] The target species data is sent to the experts whose sending times do not exceed a first preset threshold during the experts' free time.

[0123] In one embodiment, obtaining the corresponding target classification level according to the recognition result includes:

[0124] In the case where the recognition result is a classification element of species or below, a classification element to be selected is selected from the recognition result according to the confidence, user instruction or the classification element of the recognition result, and the classification element to which the classification element to be selected belongs is used as the target classification element;

[0125] When the recognition result is a genus or above classification element, a candidate classification element is selected from the recognition result according to the confidence, user instructions or the classification element of the recognition result, and the candidate classification element or the classification element to which it belongs is used as the target classification element.

[0126] In one embodiment, a computer readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the following steps are implemented:

[0127] Acquire target data of a target species, wherein the target data includes: target species data and target location data;

[0128] Input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result;

[0129] Filtering a corresponding target expert list from an expert database according to the target classification level or target location data;

[0130] Sending the target species data to experts in the target expert list;

[0131] Receive the evaluation feedback from the experts in the target expert list, and determine the optimal evaluation feedback according to the evaluation feedback from each expert in the target expert list.

[0132] In one embodiment, the expert database is a database storing expert information on various species identification aspects, wherein the expert information includes: expert ID, identification field to which the expert belongs, taxonomic group to which the expert belongs, and geographical area of ​​concern to which the expert focuses; the corresponding target expert list is selected from the expert database according to the target taxonomic level or target location data, including:

[0133] Selecting experts whose taxonomic groups include target taxonomic elements from the expert database to obtain a first expert list;

[0134] Selecting experts whose taxonomic group includes a first identification taxonomic class from the expert database to obtain a second expert list, wherein the first identification taxonomic class is the taxonomic class to which the target taxonomic class belongs;

[0135] Selecting experts whose geographical area of ​​interest includes the target location data and whose taxonomic group includes the second identification taxonomic category from the expert database to obtain a third expert list, wherein the second identification taxonomic category is the taxonomic category to which the first identification taxonomic category belongs;

[0136] A target expert list is selected from the first expert list, the second expert list and the third expert list according to the first preset instruction.

[0137] In one embodiment, selecting a target expert list from the first expert list, the second expert list, and the third expert list according to the first preset instruction includes:

[0138] Determine whether the number in the first expert list reaches a preset recommended number;

[0139] If yes, selecting a preset recommended number of experts from the first expert list as a target expert list;

[0140] If not, determine whether the sum of the number of experts in the second expert list reaches the preset recommended number. If so, select all the experts in the first expert list and the remaining experts in the second expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, determine whether the number of experts in the second expert list and the third expert list reaches the preset recommended number. If so, select all the experts in the second expert list and the remaining experts in the third expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, use all the experts in the second expert list and the third expert list as the target expert list.

[0141] In one embodiment, after selecting a corresponding target expert list from an expert database according to the target classification level or target location data, the following steps are performed:

[0142] The experts in the target expert list are sorted according to preset priorities, wherein the preset priorities include: expert feedback probability or expert professional title level.

[0143] In one embodiment, determining the optimal evaluation feedback according to the evaluation feedback of the experts in the target expert list includes:

[0144] Counting the number of identical identification feedbacks, and taking the identification feedback with the largest number as the first identification feedback;

[0145] Comparing the relationship between each of the identification feedbacks and the corresponding expert's taxonomic group, and taking the identification feedback with the closest relationship to the corresponding expert's taxonomic group as the second identification feedback;

[0146] Compare the preset expert weights corresponding to the respective appraisal feedbacks, and take the appraisal feedback with the highest preset expert weight as the third appraisal feedback;

[0147] The optimal identification feedback is selected from the first identification feedback, the second identification feedback and the third identification feedback according to the second preset instruction.

[0148] In one embodiment, sending the target species data to an expert in a target expert list includes:

[0149] Counting the number of times a message is sent to each expert in the target expert list within a preset time;

[0150] The target species data is sent to the experts whose sending times do not exceed a first preset threshold during the experts' free time.

[0151] In one embodiment, obtaining the corresponding target classification level according to the recognition result includes:

[0152] In the case where the recognition result is a classification element of species or below, a classification element to be selected is selected from the recognition result according to the confidence, user instruction or the classification element of the recognition result, and the classification element to which the classification element to be selected belongs is used as the target classification element;

[0153] When the recognition result is a genus or above classification element, a candidate classification element is selected from the recognition result according to the confidence, user instructions or the classification element of the recognition result, and the candidate classification element or the classification element to which it belongs is used as the target classification element.

[0154] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0155] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. A species identification expert docking method based on AI recognition, characterized in that: include: Acquire target data of a target species, wherein the target data includes: target species data and target location data; Input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result; Filtering a corresponding target expert list from an expert database according to the target classification level or target location data; Sending the target species data to experts in the target expert list; Receiving evaluation feedback from experts in the target expert list, and determining optimal evaluation feedback according to the evaluation feedback from each expert in the target expert list; The expert database is a database storing expert information on various species identification aspects, wherein the expert information includes: expert ID, identification field to which the expert belongs, taxonomic group to which the expert belongs, and geographical area of ​​concern to which the expert focuses; the corresponding target expert list is selected from the expert database according to the target taxonomic class or target location data, including: Selecting experts whose taxonomic groups include target taxonomic elements from the expert database to obtain a first expert list; Selecting experts whose taxonomic group includes a first identification taxonomic class from the expert database to obtain a second expert list, wherein the first identification taxonomic class is the taxonomic class to which the target taxonomic class belongs; Selecting experts whose geographical area of ​​interest includes the target location data and whose taxonomic group includes the second identification taxonomic level from the expert database to obtain a third expert list, wherein the second identification taxonomic level is the taxonomic level to which the first identification taxonomic level belongs; and the highest level that can be set for the second identification taxonomic level is limited according to the characteristics of the expert group of each biological taxonomic group; Selecting a target expert list from the first expert list, the second expert list, and the third expert list according to a first preset instruction; The step of selecting a target expert list from the first expert list, the second expert list, and the third expert list according to the first preset instruction includes: Determine whether the number in the first expert list reaches a preset recommended number; If yes, selecting a preset recommended number of experts from the first expert list as a target expert list; If not, determine whether the sum of the number of experts in the second expert list reaches the preset recommended number. If so, select all the experts in the first expert list and the remaining experts in the second expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, determine whether the number of experts in the second expert list and the third expert list reaches the preset recommended number. If so, select all the experts in the second expert list and the remaining experts in the third expert list to make the number of experts reach the preset recommended number, and use the selected experts as the target expert list. If not, use all the experts in the second expert list and the third expert list as the target expert list.

2. The method for connecting species identification experts based on AI recognition as claimed in claim 1, characterized in that: After selecting a corresponding target expert list from an expert database according to the target classification level or target location data, the method further comprises: Experts in the target expert list are sorted according to preset priorities, and a preset number of target experts are selected from the front to the back according to the sorting in the target expert list, wherein the preset priorities include: probability of expert feedback, expert professional title level, or probability of users giving favorable comments on expert historical appraisals.

3. The method for connecting species identification experts based on AI recognition as claimed in claim 1, characterized in that: The determining of the optimal evaluation feedback according to the evaluation feedback of the experts in the target expert list includes: Counting the number of identical identification feedbacks, and taking the identification feedback with the largest number as the first identification feedback; Comparing the relationship between each of the identification feedbacks and the corresponding expert's taxonomic group, and taking the identification feedback with the closest relationship to the corresponding expert's taxonomic group as the second identification feedback; Compare the preset expert weights corresponding to the respective appraisal feedbacks, and take the appraisal feedback with the highest preset expert weight as the third appraisal feedback; The optimal identification feedback is selected from the first identification feedback, the second identification feedback and the third identification feedback according to the second preset instruction.

4. The method for connecting species identification experts based on AI recognition as claimed in claim 1, characterized in that: The step of sending the target species data to experts in a target expert list includes: Counting the number of times a message is sent to each expert in the target expert list within a preset time; The target species data is sent to the experts whose sending times do not exceed a first preset threshold during the experts' free time.

5. The method for connecting species identification experts based on AI recognition as claimed in claim 1, characterized in that: The step of obtaining the corresponding target classification level according to the recognition result includes: In the case where the recognition result is a classification element of species or below, a classification element to be selected is selected from the recognition result according to the confidence, user instruction or the classification element of the recognition result, and the classification element to which the classification element to be selected belongs is used as the target classification element; When the recognition result is a genus or above classification element, a candidate classification element is selected from the recognition result according to the confidence, user instructions or the classification element of the recognition result, and the candidate classification element or the classification element to which it belongs is used as the target classification element.

6. The species identification expert docking system based on AI recognition is characterized by: include: A data acquisition module, used to acquire target data of a target species, wherein the target data includes: target species data and target location data; A species recognition module is used to input the target species data into a trained AI recognition model to obtain a recognition result, and obtain a corresponding target classification level according to the recognition result; An expert screening module is used to screen out a corresponding target expert list from an expert database according to the target taxonomic class or target location data; the expert database is a database storing expert information on various species identification aspects, the expert information including: expert ID, identification field to which the expert belongs, taxonomic group to which the expert belongs, and geographical area of ​​concern to which the expert is concerned; the method of screening out a corresponding target expert list from the expert database according to the target taxonomic class or target location data includes: Selecting experts whose taxonomic groups include target taxonomic elements from the expert database to obtain a first expert list; Selecting experts whose taxonomic group includes a first identification taxonomic class from the expert database to obtain a second expert list, wherein the first identification taxonomic class is the taxonomic class to which the target taxonomic class belongs; Selecting experts whose geographical area of ​​interest includes the target location data and whose taxonomic group includes the second identification taxonomic level from the expert database to obtain a third expert list, wherein the second identification taxonomic level is the taxonomic level to which the first identification taxonomic level belongs; and the highest level that can be set for the second identification taxonomic level is limited according to the characteristics of the expert group of each biological taxonomic group; Selecting a target expert list from the first expert list, the second expert list, and the third expert list according to a first preset instruction; The step of selecting a target expert list from the first expert list, the second expert list, and the third expert list according to the first preset instruction includes: Determine whether the number in the first expert list reaches a preset recommended number; If yes, selecting a preset recommended number of experts from the first expert list as a target expert list; If not, determine whether the sum of the number of experts in the second expert list reaches the preset recommended number. If so, select all the experts in the first expert list and the remaining experts in the second expert list so that the number of experts reaches the preset recommended number, and use the selected experts as the target expert list. If not, determine whether the number of experts in the second expert list and the third expert list reaches the preset recommended number. If so, select all the experts in the second expert list and the remaining experts in the third expert list so that the number of experts reaches the preset recommended number, and use the selected experts as the target expert list. If not, use all the experts in the second expert list and the third expert list as the target expert list; A data sending module, used for sending the target species data to experts in the target expert list; The feedback receiving and determining module is used to receive the appraisal feedback from the experts in the target expert list, and determine the optimal appraisal feedback according to the appraisal feedback from each expert in the target expert list.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.