A method, system, device and storage medium for improving bird category recognition accuracy

Abstract

The present invention is applicable to the field of sound recognition technology, and provides a method, system, device and storage medium for improving the accuracy of bird category recognition, obtaining bird sound or image data; obtaining recognition results based on a bird recognition model; filtering the recognition results based on a white list, a black list and a gray list to obtain processing measures; wherein the processing measures include: if the recognition result meets the white list, outputting the recognition result as a bird category; if the recognition result meets the black list, collecting the recognition results, and adding the recognition results to the training set of the recognition model; if the recognition result meets the gray list, introducing expert knowledge to confirm the recognition result, and adding the recognition result to the white list; this application improves the accuracy of bird sound recognition through the setting of white, black and gray lists, and iterates the model recognition capability in a targeted and rapid manner.

Description

Technical Field

[0001] The present invention belongs to the technical field of sound recognition, and in particular relates to a method, system, device and storage medium for improving the accuracy of bird category recognition. Background Art

[0002] In terms of wetland protection, long-term monitoring and recording of birds in nature are carried out to identify bird species in order to study the survival and migration status of species and take corresponding measures to protect endangered species.

[0003] Currently, most bird monitoring solutions in smart wetlands are divided into two categories: image recognition and audio recognition. The main solution for audio recognition is to install a large number of recorders in the protected area to collect audio data from birds in the protected area. The collected data is then segmented and manually labeled as training samples. The AI ​​model learns some important features of bird calls through the existing data, including but not limited to frequency, loudness and other information. The classifier then analyzes and calculates the audio Mel spectrum to determine the specific type of bird call.

[0004] It can be seen that in the above process, due to the complex wild environment, large background noise and certain similarities in the calls of some birds, the recognition accuracy of the recognition model is not high in some specific scenarios. Summary of the Invention

[0005] The purpose of the embodiments of the present application is to provide a method for improving the accuracy of bird category recognition, aiming to solve the problem of low accuracy of bird category recognition models in the prior art.

[0006] The embodiment of the present application is implemented as follows: a method for improving the accuracy of bird category recognition, the method comprising:

[0007] Acquiring bird sound or image data;

[0008] According to the bird recognition model, the recognition result is obtained;

[0009] Filter the identification results based on the whitelist, blacklist, and graylist to obtain treatment measures;

[0010] The processing measures include: if the recognition result matches the whitelist, outputting the recognition result as a bird category; if the recognition result matches the blacklist, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result matches the graylist, introducing expert knowledge to confirm the recognition result and adding the recognition result to the whitelist;

[0011] The whitelist is the bird categories that are determined to exist in the area within a unit time, the blacklist is the bird categories that are determined not to exist in the area within a unit time, and the graylist is the bird categories in the area within a unit time excluding the whitelist and blacklist.

[0012] Another object of the present invention is to provide a system for improving the accuracy of bird classification recognition, the system comprising:

[0013] A data acquisition module, used for acquiring bird sound or image data;

[0014] A bird recognition module is used to obtain recognition results based on a bird recognition model;

[0015] The result screening module is used to screen the recognition results based on a whitelist, a blacklist and a graylist to obtain processing measures; wherein the processing measures include: if the recognition result meets the whitelist, outputting the recognition result as a bird category; if the recognition result meets the blacklist, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result meets the graylist, introducing expert knowledge to confirm the recognition result, and adding the recognition result to the whitelist.

[0016] Another object of an embodiment of the present application is a computer device, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the method for improving the accuracy of bird category identification.

[0017] Another object of an embodiment of the present application is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the processor executes the steps of the method for improving the accuracy of bird category recognition.

[0018] An embodiment of the present application provides a method for improving the accuracy of bird category recognition. Through a bird recognition model, bird category recognition results are obtained, and the recognition results are screened through a white list, a black list, and a gray list. The erroneous recognition results filtered out by the black list can be used to construct an erroneous sample data set, added to the recognition model training set, and the recognition model can be quickly iterated. Through the gray list and the introduction of expert knowledge, new bird species that appear in this ecological area per unit time can be supplemented, and erroneous recognition results can be quickly screened out, thereby improving the accuracy of the recognition results. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 Flowchart of a method for improving bird category recognition accuracy provided in an embodiment of the present application;

[0020] Figure 2A flowchart of screening white, black, and gray lists for bird category identification provided in an embodiment of the present application;

[0021] Figure 3 Expert identification judgment diagram provided by the embodiment of this application;

[0022] Figure 4 This is a structural block diagram of a device for improving bird category recognition accuracy provided in an embodiment of the present application;

[0023] Figure 5 FIG. 1 is a block diagram of the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0025] It is understood that the terms "first," "second," etc., used herein may be used to describe various elements, but unless otherwise specified, these elements are not limited by these terms. These terms are only used to distinguish a first element from another element. For example, a first xx script may be referred to as a second xx script, and similarly, a second xx script may be referred to as a first xx script without departing from the scope of this application.

[0026] like Figure 1 As shown, in one embodiment, a method for improving the accuracy of bird category recognition is proposed, which may specifically include the following steps:

[0027] Step S102, obtaining bird sound or image data;

[0028] Step S104, obtaining a recognition result according to the bird recognition model;

[0029] Step S106, filtering the recognition results based on the whitelist, blacklist, and graylist to obtain processing measures; wherein the processing measures include: if the recognition result meets the whitelist, outputting the recognition result as a bird category; if the recognition result meets the blacklist, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result meets the graylist, introducing expert knowledge to confirm the recognition result, and adding the recognition result to the whitelist.

[0030] In this embodiment, the image or sound data of birds can be used as the basis for identification. The bird recognition model can be an existing image recognition model or a sound recognition model, for example, the YOLO series target detection model, which is not specifically limited. The white list is the bird category determined to exist in the area within a unit time, the black list is the bird category determined not to exist in the area within a unit time, and the gray list is the bird category in the area excluding the white list and the black list within a unit time. The unit time can be in months or years. The area is a relatively fixed area, which can be a wetland protection area or a city park. The bird categories in the white list and the black list can be obtained based on field surveys of the area or based on monitoring data of the area over the years. Screening is to compare the recognition results with the white list, black list and gray list in turn. If the recognition result appears in the corresponding list, the operation of the corresponding list is executed. For example, bird sound data is obtained, the category of the bird is identified through the sound data, and the corresponding bird is screened in the white list, and this recognition result is directly output as the final bird category.

[0031] In one embodiment, Figure 2 As shown, the screening of the identification results based on the whitelist, blacklist and greylist includes the following steps:

[0032] Step S202, determining whether the bird category is in the whitelist;

[0033] Step S204: when it is determined that the bird category is in the whitelist, the recognition result is output as the bird category;

[0034] Step S206, when it is determined that the bird category is not in the whitelist, determining whether the bird category is in the blacklist;

[0035] Step S208, when it is determined that the bird category is in the blacklist, collecting the recognition results and adding the recognition results to the training set of the recognition model;

[0036] Step S210, when it is determined that the bird category is not in the blacklist, determining whether the bird category is in the gray list;

[0037] Step S212: When the bird category is in the gray list, expert knowledge is introduced to confirm the recognition result, and the recognition result is added to the white list;

[0038] Step S214: When the bird category is not in the gray list, return to the bird identification model to obtain an identification result.

[0039] In this embodiment, the screening based on the whitelist, blacklist and greylist can be carried out in the following ways:

[0040] First, the recognition result is screened to see if it is on the whitelist. If it is, the recognition result is output as a bird category. This can reflect the accuracy of the recognition model used and also provide data support for the migration patterns of birds in the area. If it is not on the whitelist, the blacklist is screened. If the recognition result appears on the blacklist, it can be discarded, that is, the recognition result is abandoned. At the same time, the incorrect result can be added to the model training dataset for iterative modeling. If the recognition result is on the graylist, expert knowledge can be introduced for judgment. If the recognition result is correct, the bird category of this recognition result can be added to the whitelist. If the recognition result is not on the graylist, the original bird image and sound data can be combined to jointly determine the bird category based on the recognition results of both aspects. When the bird category of the recognition result is not on the whitelist, blacklist, or graylist, the bird recognition model can be returned to for a new recognition, and the new recognition result can be re-judged on the whitelist, blacklist, and graylist.

[0041] In one embodiment, Figure 3 As shown, the introduction of expert knowledge to confirm the recognition result and transferring the recognition result to the whitelist includes the following steps:

[0042] Step S302, determining whether the recognition result is correct;

[0043] Step S304: When the recognition result is correct, output the recognition result and add the recognition result to the whitelist;

[0044] Step S306: When the recognition result is incorrect, the recognition result is added to the training set of the recognition model.

[0045] In this embodiment, expert knowledge can be a bird atlas or the conclusions of professionals. Since common birds in the same area will change accordingly over time, the gray list, black list and white list in this application can be adjusted according to the changes in bird categories within a unit time range in a specified area. For example, a certain area is mainly wintering water birds in winter and mainly resident birds in summer. In this way, the white list in winter includes wintering water birds, and the black list includes summer resident birds. The gray list can be birds in other areas that are not on the white list and black list, that is, birds that may appear in this area, or birds on the white list in autumn. For example, the bird categories on the gray list in December include the bird categories on the white list in November.

[0046] In one embodiment, the white list can be determined by generating a list of bird categories that appear in the area per unit time based on the area address and ecological environment. The black list can be determined by generating a list of bird categories that do not appear in the area per unit time based on the area address and ecological environment. The gray list can be determined by generating a list of bird categories that are not in the white list and the black list. The birds in the above lists are all confined to a certain area, so the migration patterns of birds in a region can be well recorded and tracked. The migration of birds in and out of this area can be observed through the gray list, and new species of birds that may appear and migrate in this area can be supplemented.

[0047] like Figure 4 As shown, in one embodiment, a system for improving the accuracy of bird category recognition is provided, the system comprising:

[0048] A data acquisition module 410 is used to acquire bird sound or image data;

[0049] The bird recognition module 420 is used to obtain a recognition result based on the bird recognition model;

[0050] The result screening module 430 is used to screen the recognition results based on a whitelist, a blacklist, and a graylist; if the recognition result meets the whitelist, the recognition result is output as a bird category; if the recognition result meets the blacklist, the recognition results are collected and added to the training set of the recognition model; if the recognition result meets the graylist, expert knowledge is introduced to confirm the recognition result, and the recognition result is added to the whitelist.

[0051] In this embodiment, the recognition accuracy is improved in real time through the blacklist-based error result filtering strategy and the graylist to whitelist conversion strategy. By establishing an error recognition result dataset and adding the error recognition dataset to the training set of the recognition model, the model accuracy is improved.

[0052] In one embodiment, the result screening module 510 specifically includes:

[0053] A white, black and grey list judgment module is used to judge whether the bird category is in the white list; when it is judged that the bird category is in the white list, the recognition result is output as the bird category;

[0054] When it is determined that the bird category is not in the whitelist, determining whether the bird category is in the blacklist;

[0055] When it is determined that the bird category is in the blacklist, collecting recognition results and adding the recognition results to the training set of the recognition model;

[0056] When it is determined that the bird category is not in the blacklist, determining whether the bird category is in the gray list;

[0057] When the bird category is in the gray list, expert knowledge is introduced to confirm the recognition result, and the recognition result is added to the white list;

[0058] When the bird category is not in the gray list, the method returns to the bird identification model to obtain an identification result.

[0059] Figure 5 FIG. 1 shows an internal structure diagram of a computer device in one embodiment. Figure 5 As shown, the computer device includes a processor, a memory, a network interface, an input device and a display screen connected via a system bus. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program, which, when executed by the processor, enables the processor to implement a method for improving the accuracy of bird category identification. The internal memory may also store a computer program, which, when executed by the processor, enables the processor to execute a method for improving the accuracy of bird category identification. The display screen of the computer device may be a liquid crystal display or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a button, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0060] Those skilled in the art will understand that Figure 5 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 shown in the figure, or combine certain components, or have a different component arrangement.

[0061] In one embodiment, the bird category recognition accuracy improvement system provided by the present application can be implemented in the form of a computer program. Figure 5 The computer device shown in FIG. 1 is run on the computer device shown in FIG. The memory of the computer device can store various program modules that constitute the bird category identification accuracy improvement system, such as: Figure 4 The data acquisition module, bird identification module and result screening module shown in the figure are computer programs composed of various program modules, which enable the processor to execute the steps of the bird category identification accuracy improvement method of each embodiment of the present application described in this specification.

[0062] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the following steps are performed:

[0063] Acquiring bird sound or image data;

[0064] According to the bird recognition model, the recognition result is obtained;

[0065] Filter the identification results based on the whitelist, blacklist, and graylist to obtain treatment measures;

[0066] Among them, the processing measures include: if the recognition result meets the white list, outputting the recognition result as a bird category; if the recognition result meets the black list, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result meets the gray list, introducing expert knowledge to confirm the recognition result, and adding the recognition result to the white list.

[0067] In one embodiment, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor performs the following steps:

[0068] Acquiring bird sound or image data;

[0069] According to the bird recognition model, the recognition result is obtained;

[0070] Filter the identification results based on the whitelist, blacklist, and graylist to obtain treatment measures;

[0071] Among them, the processing measures include: if the recognition result meets the white list, outputting the recognition result as a bird category; if the recognition result meets the black list, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result meets the gray list, introducing expert knowledge to confirm the recognition result, and adding the recognition result to the white list.

[0072] It should be understood that, although each step in the flow chart of each embodiment of the present application is shown in sequence according to the indication of the arrow, these steps are not necessarily performed in sequence according to the order indicated by the arrow. Unless there is clear explanation in this article, the execution of these steps does not have strict order restriction, and these steps can be performed in other orders. Moreover, at least a portion of the steps in each embodiment may include a plurality of sub-steps or a plurality of stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily performed in sequence, but can be performed in turn or alternately with at least a portion of other steps or sub-steps or stages of other steps.

[0073] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a non-volatile computer-readable storage medium. When executed, the program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct RAMbus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0074] The technical features of the above-mentioned embodiments can be combined arbitrarily. In order to make the description concise, not all possible combinations of the technical features in the above-mentioned 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.

[0075] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for improving the accuracy of bird category recognition, characterized in that: The method comprises: Acquiring bird sound or image data; According to the bird recognition model, the recognition result is obtained; Filter the identification results based on the whitelist, blacklist, and graylist to obtain treatment measures; The processing measures include: if the recognition result matches the whitelist, outputting the recognition result as a bird category; if the recognition result matches the blacklist, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result matches the graylist, introducing expert knowledge to confirm the recognition result and adding the recognition result to the whitelist; The whitelist is the bird categories that are determined to exist in the area within a unit time, the blacklist is the bird categories that are determined not to exist in the area within a unit time, and the graylist is the bird categories in the area within a unit time excluding the whitelist and blacklist.

2. The method for improving the accuracy of bird category recognition according to claim 1, characterized in that: The method of screening the identification results based on the whitelist, blacklist and greylist to obtain treatment measures includes the following steps: Determine whether the bird category is in the whitelist; When it is determined that the bird category is in the whitelist, outputting the recognition result as the bird category; When it is determined that the bird category is not in the whitelist, determining whether the bird category is in the blacklist; When it is determined that the bird category is in the blacklist, collecting recognition results and adding the recognition results to the training set of the recognition model; When it is determined that the bird category is not in the blacklist, determining whether the bird category is in the gray list; When the bird category is in the gray list, expert knowledge is introduced to confirm the recognition result, and the recognition result is added to the white list; When the bird category is not in the gray list, the method returns to the bird identification model to obtain an identification result.

3. The method for improving the accuracy of bird category recognition according to claim 2, characterized in that: The step of introducing expert knowledge to confirm the recognition result and transferring the recognition result to the whitelist comprises the following steps: Determining whether the recognition result is correct; When the recognition result is correct, output the recognition result and add the recognition result to the whitelist; When the recognition result is incorrect, the recognition result is added to the training set of the recognition model.

4. The method for improving the accuracy of bird category recognition according to claim 1, characterized in that: The whitelist is determined as follows: According to the regional address and ecological environment, a list of bird categories that appear in the region per unit time is generated as a whitelist.

5. The method for improving the accuracy of bird category recognition according to claim 1, characterized in that: The blacklist is determined as follows: According to the regional address and ecological environment, a list of bird categories that do not appear in the region per unit time is generated as a blacklist.

6. The method for improving the accuracy of bird category recognition according to claim 1, characterized in that: The grey list is determined as follows: A list of bird categories that are not on the white list and the black list is referred to as a grey list.

7. A bird category recognition accuracy improvement system, characterized by: The system comprises: A data acquisition module, used for acquiring bird sound or image data; A bird recognition module is used to obtain recognition results based on a bird recognition model; a result screening module for screening the recognition results based on a whitelist, a blacklist, and a graylist to obtain treatment measures; wherein the treatment measures include: if the recognition result matches the whitelist, outputting the recognition result as a bird category; if the recognition result matches the blacklist, collecting the recognition results and adding the recognition results to the training set of the recognition model; if the recognition result matches the graylist, introducing expert knowledge to confirm the recognition result and adding the recognition result to the whitelist; The whitelist is the bird categories that are determined to exist in the area within a unit time, the blacklist is the bird categories that are determined not to exist in the area within a unit time, and the graylist is the bird categories in the area within a unit time excluding the whitelist and blacklist.

8. The system for improving the accuracy of bird classification recognition according to claim 7, characterized in that: The result screening module also includes: The white, black and grey list judging module is used to judge whether the bird category is in the white list; when it is judged that the bird category is in the white list, the recognition result is output as the bird category.

9. A computer device, characterized in that: The invention comprises a memory and a processor, wherein a computer program is stored in the memory, and when the computer program is executed by the processor, the processor executes the steps of the method for improving the accuracy of bird category recognition according to any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method for improving the accuracy of bird category recognition according to any one of claims 1 to 6.

Images