Seed variety identification method and device, electronic equipment and storage medium

Abstract

The invention provides a seed variety identification method and device, electronic equipment and a storage medium, and the method comprises the steps: correcting the spectral information of a to-be-identified seed in target seeds based on the number of varieties included in the target seeds, the spectral information of the target seeds and the variety information of a fourth sample seed in the target seeds, and after obtaining the corrected spectral information of the to-be-identified seeds, inputting the corrected spectral information of the to-be-identified seeds into the seed variety identification model, and obtaining the variety information of the to-be-identified seeds output by the seed variety identification model. According to the seed variety identification method and device, the electronic equipment and the storage medium provided by the invention, on the basis of not reducing the accuracy of identifying the seed variety with the same ecological factor as the historical sample seed, the accuracy of identifying the seed variety with the different ecological factors from the historical sample seed can be improved, the seed variety identification efficiency can be improved, and the seed variety identification accuracy is improved. And a more accurate data basis can be provided for seed breeding, seed trade and agricultural production.

Description

Technical Field

[0001] The present invention relates to the field of smart agricultural technology, and in particular to a seed variety identification method, device, electronic equipment and storage medium. Background Art

[0002] Variety (genotype) is an important indicator of seed quality, significantly affecting the external morphology, internal chemical composition, vitality level, yield potential, stress resistance and other multi-dimensional characteristics of seeds. With the continuous advancement of agricultural technology and the increasing openness of the global market, the seed industry has become highly commercialized and specialized. Accurate and efficient identification of seed varieties is of great significance for ensuring the safety and stability of agricultural production, improving crop yield and quality, promoting the healthy development of the seed industry, and meeting the global market's demand for high-quality agricultural products.

[0003] In related technologies, deep learning technology can be used to identify seed varieties. However, the phenotypic characteristics of seeds (such as smoothness, particle size, weight, and density) are greatly affected by ecological factors such as seed storage time, cultivation year, and cultivation conditions. The heterogeneity of sample seeds used for model training and seeds to be identified in terms of ecological factors will seriously affect the seed variety identification accuracy of the seed variety identification model constructed and trained based on deep learning calculations in actual applications.

[0004] In the related art, the seed variety recognition model can be trained with the newly added sample data, so that the seed variety recognition model can obtain the ability to identify the varieties of seeds with the same ecological factors as the newly added sample seeds. However, after the seed variety recognition model is trained with the newly added sample data, the information entropy of the sample data will increase, thereby reducing the accuracy of the seed variety recognition model in identifying the seed varieties with the same ecological factors as the historical sample seeds. Therefore, how to improve the accuracy of identifying seed varieties with different ecological factors from the historical sample seeds without reducing the accuracy of identifying seed varieties with the same ecological factors as the historical sample seeds is a technical problem that needs to be solved urgently in this field. Summary of the invention

[0005] The present invention provides a seed variety identification method, device, electronic device and storage medium, which are used to solve the problem that the heterogeneity of ecological factors between sample seeds used for model training and seeds to be identified in the prior art leads to low seed variety identification accuracy of seed variety identification models constructed and trained based on deep learning calculations in practical applications. The method improves the accuracy of identifying seed varieties with different ecological factors from historical sample seeds without reducing the accuracy of identifying seed varieties with the same ecological factors as historical sample seeds.

[0006] The invention provides a seed variety identification method, comprising the following steps.

[0007] Obtain the spectral information of the target seeds and the variety information of the fourth sample seeds in the target seeds, determine the variety included in the target seeds as the target variety, obtain the quantity of each target variety as the first quantity, the varieties included in the fourth sample seeds are each target variety, and the target seeds have the same ecological factors, where the ecological factors include factors affecting seed cultivation and / or seed storage; Based on the first quantity, and the spectral information and variety information of the fourth sample seeds, correct the spectral information of the seeds to be identified in the target seeds, and obtain the corrected spectral information of the seeds to be identified, where the variety of the seeds to be identified is one of each target variety; Input the corrected spectral information of the seeds to be identified into the variety identification model, and obtain the variety information of the seeds to be identified output by the variety identification model; Wherein, the variety identification model is obtained by training based on the spectral information and variety information of the first sample seeds; the varieties of the first sample seeds include each alternative variety corresponding to the seeds to be identified.

[0008] According to a seed variety identification method provided by the present invention, the step of correcting the spectral information of the seeds to be identified in the target seeds based on the first quantity, and the spectral information and variety information of the fourth sample seeds, and obtaining the corrected spectral information of the seeds to be identified includes: When it is determined that the first quantity is not greater than the quantity threshold, based on the variety information of the first sample seeds, determine the first sample seeds with the variety being the target variety as the second sample seeds; For each target variety, randomly select a second quantity from the second sample seeds of each target variety as the third sample seeds; Based on the spectral information and variety information of the third sample seeds, and the spectral information and variety information of the fourth sample seeds, generate a transfer matrix corresponding to the seeds to be identified by means of spectral calibration transfer; Based on the transfer matrix corresponding to the seeds to be identified, correct the spectral information of the seeds to be identified, and obtain the corrected spectral information of the seeds to be identified.

[0009] According to a seed variety identification method provided by the present invention, the step of correcting the spectral information of the seeds to be identified in the target seeds based on the first quantity, and the spectral information and variety information of the fourth sample seeds, and obtaining the corrected spectral information of the seeds to be identified includes: When it is determined that the first quantity is greater than the quantity threshold, cluster the spectral information of the target seeds, and divide the target seeds into multiple seed groups according to the clustering result; For each seed group, generate a transfer matrix corresponding to each seed group based on the spectral information and variety information of the fourth sample seeds in each seed group; Based on the transfer matrix corresponding to each seed group, modify the spectral information of the seeds to be identified in each seed group to obtain the corrected spectral information of the seeds to be identified in each seed group; Stitch the corrected spectral information of the seeds to be identified in each seed group to obtain the corrected spectral information of the seeds to be identified.

[0010] According to a seed variety identification method provided by the present invention, the generating a transfer matrix corresponding to each seed group based on the spectral information and variety information of the fourth sample seeds in each seed group includes: Based on the variety information of the fourth sample seeds in each seed group, determine the first sample seeds with the same variety as the fourth sample seeds in each seed group as the fifth sample seeds corresponding to each seed group; Randomly select a third quantity from the fifth initial sample seeds corresponding to each seed group as the sixth sample seeds corresponding to each seed group; Based on the spectral information and variety information of the sixth sample seeds corresponding to each seed group, and the spectral information and variety information of the fourth sample seeds in each seed group, generate a transfer matrix corresponding to each seed group by means of spectral calibration transfer.

[0011] According to a seed variety identification method provided by the present invention, the generating a transfer matrix corresponding to the seeds to be identified by means of spectral calibration transfer based on the spectral information and variety information of the third sample seeds and the spectral information and variety information of the fourth sample seeds includes: Based on the spectral information and variety information of the third sample seeds and the spectral information and variety information of the fourth sample seeds, generate a transfer matrix corresponding to the seeds to be identified by means of a direct standardization algorithm.

[0012] According to a seed variety identification method provided by the present invention, the clustering of the spectral information of the target seeds includes: Perform distance measurement on the spectral information of the target seeds based on the K-means clustering algorithm to obtain the clustering result.

[0013] The present invention also provides a seed variety identification device, including the following modules: An information acquisition module, configured to acquire the spectral information of target seeds and the variety information of the fourth sample seeds in the target seeds, determine the variety included in the target seeds as the target variety, acquire the quantity of each target variety as the first quantity, the varieties included in the fourth sample seeds are each target variety, and the target seeds have the same ecological factors, where the ecological factors include factors affecting seed cultivation and / or seed storage; A spectral correction module, configured to correct the spectral information of the seeds to be identified in the target seeds based on the first quantity, the spectral information and variety information of the fourth sample seeds, and acquire the corrected spectral information of the seeds to be identified, where the variety of the seeds to be identified is one of each target variety; A variety identification module, configured to input the corrected spectral information of the seeds to be identified into a variety identification model, and acquire the variety information of the seeds to be identified output by the variety identification model; Wherein, the variety identification model is obtained by training based on the spectral information of the first sample seeds and the variety information of the first sample seeds; the varieties of the first sample seeds include each alternative variety corresponding to the seeds to be identified.

[0014] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes the computer program, the seed variety identification method as described in any one of the above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the seed variety identification method as described in any one of the above is implemented.

[0016] The present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, the seed variety identification method as described in any one of the above is implemented.

[0017] The seed variety identification method, device, electronic device and storage medium provided by the present invention correct the spectral information of the seeds to be identified in the target seeds based on the number of varieties included in the target seeds, the spectral information of the target seeds, and the variety information of the fourth sample seeds in the target seeds. After obtaining the corrected spectral information of the seeds to be identified, the corrected spectral information of the seeds to be identified is input into the seed variety identification model, and the variety information of the seeds to be identified output by the seed variety identification model is obtained. It can correct the spectral information of a large number of seeds to be identified with unknown varieties in the target seeds based on the spectral information of the fourth sample seeds of a small number of known varieties in the target seeds and the spectral information of the first sample seeds of known varieties used for training the seed variety identification model. By correcting the spectral information of the seeds to be identified, the influence of the heterogeneity of ecological factors on the spectral information of the seeds is corrected, so that the accuracy of identifying the seed varieties with ecological factors different from those of the historical sample seeds can be improved on the basis of not reducing the accuracy of identifying the seed varieties with the same ecological factors as the historical sample seeds, the efficiency of seed variety identification can be improved, a more accurate data basis can be provided for seed breeding, seed trade and agricultural production, and the safety and stability of agricultural production can be better guaranteed, the yield and quality of crops can be improved, the healthy development of the seed industry can be promoted, and the global market demand for high-quality agricultural products can be met. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0019] Figure 1 is one of the flow diagrams of the seed variety identification method provided by the present invention.

[0020] Figure 2 is the second flow diagram of the seed variety identification method provided by the present invention.

[0021] Figure 3 is the structural diagram of the seed variety identification device provided by the present invention.

[0022] Figure 4 is the structural diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the present invention. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts shall fall within the protection scope of the present invention.

[0024] In the description of the invention, it should be noted that unless otherwise clearly defined and limited, the terms "installation", "connection" and "coupling" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection or an integral connection; it may be a mechanical connection or an electrical connection; it may be a direct connection or an indirect connection through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0025] In the description of this application, the terms "first", "second", etc. are used to distinguish similar objects, rather than to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances so that the embodiments of this application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are usually of the same kind, and the number of objects is not limited. For example, the first object can be one or more. In addition, in the description of this application, "and / or" means at least one of the connected objects, and the character " / " generally means that the related objects before and after are in an "or" relationship.

[0026] It should be noted that with the high degree of commercialization and specialization of the seed industry, there has emerged in the seed market the sale of seeds of poor-quality varieties that are unsalable by posing as seeds of high-quality varieties that are in high demand, in order to obtain improper benefits. Such acts not only infringe on the variety rights of breeders, but also directly cause huge economic losses to manufacturers, distributors and terminal farmers in the seed supply chain. In the long run, they also pose a potential threat to food security. Therefore, it is of great significance to identify and authenticate seed varieties throughout the entire production, processing and sales chain of seeds for combating variety fraud and regulating the seed industry.

[0027] In the related art, traditional seed variety detection methods can be based on DNA molecular markers, high performance liquid chromatography, gas chromatography, protein electrophoresis and other methods to detect seed varieties. However, the above traditional seed variety detection methods have defects such as complicated operation, high cost, long time consumption and strong dependence on the experience of operators, and the above traditional seed variety detection methods will cause losses of experimental samples.

[0028] The traditional seed identification method in the related art can use deep learning technology to achieve non-destructive detection of seed varieties. Specifically, the traditional seed identification method can use the deep learning model to build an initial model, and then the initial model can be trained based on sample data to obtain a trained seed variety identification model for seed variety identification.

[0029] Since model construction and model training require a lot of time and labor costs, it is usually expected that the trained seed variety identification model has good robustness and transferability, and that the trained seed variety identification model can be effective for a long time and have good performance.

[0030] However, the phenotypic characteristics of seeds (such as smoothness, particle size, weight and density) are greatly affected by ecological factors such as seed storage time, cultivation year and cultivation conditions. The heterogeneity of ecological factors between the sample seeds used for model training and the seeds to be identified will seriously affect the accuracy of seed variety identification in practical applications of the seed variety identification model constructed and trained based on deep learning calculations.

[0031] In the related art, the seed variety recognition model can be trained with the newly added sample data, so that the seed variety recognition model can obtain the ability to identify the varieties of seeds with the same ecological factors as the newly added sample seeds. However, after the seed variety recognition model is trained with the newly added sample data, the information entropy of the sample data will increase, thereby reducing the accuracy of the seed variety recognition model in identifying the seed varieties with the same ecological factors as the historical sample seeds. Therefore, how to improve the accuracy of identifying seed varieties with different ecological factors from the historical sample seeds without reducing the accuracy of identifying seed varieties with the same ecological factors as the historical sample seeds is a technical problem that needs to be solved urgently in this field.

[0032] In view of this, the present invention provides a method for identifying seed varieties. The seed variety identification method provided by the present invention proposes a spectral calibration transfer strategy between batches corresponding to data between varieties in view of the problem that seed variety identification is greatly affected by the heterogeneity of ecological factors. Samples of the same variety with different ecological factors are used as calibration transfer objects, and samples of the same variety with different ecological factors are calibrated. On this basis, in order to reduce the interference of redundant information between species, a method of grouped calibration transfer after clustering is proposed. Through the calibration transfer strategy proposed by the seed variety identification method provided by the present invention, the spectral calibration of the seeds to be identified can be completed, so that the spectral information characteristics of the seeds to be identified are closer to the spectral information characteristics of the modeled seeds, thereby improving the variety identification ability of the trained seed variety identification model for seeds with different ecological factors. The seed variety identification method provided by the present invention can be used to calibrate the spectral differences of the same material in different batches and improve the identification ability of the model for samples in different batches. This makes it have very good application prospects in variety identification and purity detection in the process of seed production and circulation.

[0033] The following combines Figure 1 - Figure 2 to describe the seed variety identification method provided by the present invention.

[0034] Figure 1 is one of the schematic flowcharts of the seed variety identification method provided by the present invention. As Figure 1 shown, the method includes the following: Step 101, obtain the spectral information of the target seeds and the variety information of the fourth sample seeds in the target seeds, determine the variety included in the target seeds as the target variety, obtain the quantity of each target variety as the first quantity, the varieties included in the fourth sample seeds are each target variety, and the target seeds have the same ecological factors, where the ecological factors include factors affecting seed cultivation and / or seed storage.

[0035] It should be noted that the execution subject in the embodiments of the present invention is a seed variety identification device. The above-mentioned seed variety identification device can be configured in an electronic device such as a computer or a server.

[0036] Specifically, the seeds to be identified in the target seeds are the identification objects of the seed identification method provided by the present invention. The target seeds can be determined based on actual needs. For example, the seeds produced by the same batch of a certain seed production company can be determined as the target seeds. There is no specific limitation on the target seeds in the embodiments of the present invention.

[0037] It should be noted that the number of target seeds in the embodiments of the present invention is multiple, and all the target seeds have the same ecological factors. Among them, the ecological factors can include but are not limited to at least one of the seed storage time, the seed cultivation year, and the seed cultivation conditions.

[0038] In the embodiments of the present invention, it can be used represents the set of seeds to be identified, denoted by X u the -th spectral information of the seeds to be identified. Among them, , represents the number of seeds to be identified.

[0039] In the embodiments of the present invention, the spectral information of the seeds to be identified can be obtained in various ways. For example, based on the input of the user, the spectral information of the seeds to be identified can be obtained; or, the spectral information of the seeds to be identified sent by other electronic devices can also be received. In the embodiments of the present invention, there is no limitation on the specific way of obtaining the spectral information of the seeds to be identified.

[0040] Specifically, in the embodiments of the present invention, X t can be used to represent the -th spectral information of the target seeds. Among them, , represents the number of target seeds.

[0041] In the embodiments of the present invention, the spectral information of the target seeds can be obtained in various ways X t , for example, based on the input of the user, the spectral information of the target seeds can be obtained X t ; or, the spectral information of the target seeds sent by other electronic devices can also be received X t .

[0042] In the embodiments of the present invention, a fourth number of target seeds can be randomly selected as the fourth sample seeds.

[0043] It should be noted that the fourth number in the embodiments of the present invention can be determined based on prior knowledge and / or actual situations. For example, the fourth number can be rounded to 20%, 25% or 30% of the number of target seeds. There is no limitation on the specific value of the fourth number in the embodiments of the present invention.

[0044] In the embodiments of the present invention, can be used to represent the set of the fourth sample seeds, and is used to represent the set of target seeds, , and X l is used to represent the spectral information of the fourth sample seeds.

[0045] After determining the fourth sample seeds from the target seeds, the variety of the fourth sample seeds can be detected by traditional variety detection methods to obtain the variety information of the fourth sample seeds indicating the variety of the fourth sample seeds.

[0046] In the embodiments of the present invention, Y l is used to represent the variety information of the l -th fourth sample seed.

[0047] After determining the fourth sample seeds among the target seeds, the seeds in the target seeds except the fourth sample seeds are determined as the seeds to be identified.

[0048] It can be understood that the number of seeds to be identified can be multiple.

[0049] It should be noted that in the embodiments of the present invention, the varieties included in the target seeds can all be determined as the target varieties. For example, when the target seeds include seeds of variety A, variety B, and variety C, variety A, variety B, and variety C can all be determined as the target varieties.

[0050] It should be noted that the varieties (i.e., the target varieties) included in the target seeds in the embodiments of the present invention are known. For example, the above-mentioned seed production company has determined the variety to be produced in this batch before production. In the seed production of this batch, seeds of the variety to be produced in this batch are cultivated respectively under the same ecological factors. After completing the seed production of this batch, the seeds obtained from this batch of production can be determined as the target seeds. Then, the varieties included in the target seeds are the varieties to be produced in this batch. Correspondingly, the target variety is the variety to be produced in this batch.

[0051] Therefore, in the embodiments of the present invention, the number of target varieties included in the target seeds can be determined based on the user's input, and then the number of each target variety can be obtained as the first quantity.

[0052] Step 102: Based on the first quantity, the spectral information and variety information of the fourth sample seeds, correct the spectral information of the seeds to be identified in the target seeds, and obtain the corrected spectral information of the seeds to be identified. The variety of the seeds to be identified is one of the target varieties.

[0053] Specifically, after obtaining the first quantity, different correction strategies can be selected based on the first quantity. Then, based on the selected correction strategy, the spectral information and variety information of the fourth sample seeds, the spectral information of the seeds to be identified can be corrected by means of numerical calculation, mathematical statistics, deep learning technology, etc., to obtain the corrected spectral information of the seeds to be identified. In the embodiments of the present invention, X ut is used to represent the corrected spectral information of the ut -th seed to be identified.

[0054] Step 103: Input the corrected spectral information of the seed to be identified into the variety identification model, and obtain the variety information of the seed to be identified output by the variety identification model.

[0055] Among them, the variety identification model is obtained after being trained based on the spectral information of the first sample seeds and the variety information of the first sample seeds; the varieties of the first sample seeds include each alternative variety corresponding to the seeds to be identified.

[0056] It should be noted that the variety identification model in the embodiments of the present invention can be constructed based on a machine learning model. Using the spectral information of the first sample seeds as the training samples and the variety information in the first sample as the sample labels for training.

[0057] Optionally, the variety identification model in the embodiments of the present invention can be constructed based on the Linear Discriminant Analysis (LDA) classifier in machine learning.

[0058] In the embodiments of the present invention, can be used to represent the set of the first sample seeds, and X s is used to represent the spectral information of the first sample seeds, and Y s is used to represent the variety information of the first sample seeds.

[0059] It should be noted that the spectral information of the seed set in the embodiments of the present invention can be represented by a matrix. For example, the spectral information of the first sample seeds can be represented by a matrix of , where represents the number of spectral feature bands; represents the number of the first sample seeds; x ij represents the spectral information of the th first sample seed of the target variety.

[0060] After obtaining the corrected spectral information of the seed to be identified, the corrected spectral information of the seed to be identified can be input into the above variety identification model, and then the variety information of the seed to be identified output by the above variety identification model can be obtained.

[0061] In the embodiment of the present invention, based on the number of varieties included in the target seeds, the spectral information of the target seeds, and the variety information of the fourth sample seeds in the target seeds, the spectral information of the seeds to be identified in the target seeds is corrected. After obtaining the corrected spectral information of the seeds to be identified, the corrected spectral information of the seeds to be identified is input into the seed variety identification model, and the variety information of the seeds to be identified output by the seed variety identification model is obtained. It is possible to correct the spectral information of a large number of seeds to be identified with unknown varieties in the target seeds based on the spectral information of the fourth sample seeds with a small number of known varieties in the target seeds and the spectral information of the first sample seeds of the known varieties used for training the seed variety identification model. By correcting the spectral information of the seeds to be identified, the influence of the heterogeneity of ecological factors on the spectral information of the seeds is corrected, so that the accuracy of identifying the seed varieties with ecological factors different from those of the historical sample seeds can be improved on the basis of not reducing the accuracy of identifying the seed varieties with the same ecological factors as the historical sample seeds. The efficiency of seed variety identification can be improved, a more accurate data basis can be provided for seed breeding, seed trade, and agricultural production, and the safety and stability of agricultural production can be better guaranteed, the yield and quality of crops can be improved, the healthy development of the seed industry can be promoted, and the global market demand for high-quality agricultural products can be met.

[0062] Figure 2 is the second flow diagram of the seed variety identification method provided by the present invention. As Figure 2 shown, as an optional embodiment, based on the first quantity and the spectral information and variety information of the fourth sample seeds, the spectral information of the seeds to be identified in the target seeds is corrected to obtain the corrected spectral information of the seeds to be identified, including: when it is determined that the first quantity is not greater than the quantity threshold, based on the variety information of the first sample seeds, the first sample seeds with the variety of the target variety are determined as the second sample seeds.

[0063] Specifically, to obtain the first quantity, the first quantity can be compared with the quantity threshold.

[0064] If the first quantity is not greater than the quantity threshold, it can indicate that the number of varieties included in the target seeds is small, and the first correction strategy can be selected to correct the spectral information of the seeds to be identified.

[0065] It should be noted that in the embodiment of the present invention, different correction strategies are adopted to correct the spectral information of the seeds to be identified according to the number of varieties included in the target seeds, which can improve the efficiency of correcting the spectral information of the seeds to be identified on the basis of ensuring the correction accuracy. The above quantity threshold can be determined based on prior knowledge and / or actual situations. For example, the value of the above numerical threshold can be 5, 6, or 7. The specific value of the above quantity threshold is not limited in the embodiment of the present invention.

[0066] The specific content of the first correction strategy is as follows: Based on the variety information of the first sample seeds, the first sample seeds with the variety being the target variety in the first sample seeds are determined as the second sample seeds.

[0067] For each target variety, randomly select a second number of the second sample seeds of each target variety as the third sample seeds.

[0068] Specifically, in the embodiments of the present invention, can be used to identify the target variety, , represents the first number.

[0069] For the target variety , a second number of the second sample seeds with the variety being the target variety can be randomly selected as the third sample seeds with the variety being the target variety .

[0070] It should be noted that the second number in the embodiments of the present invention can be determined based on prior knowledge and / or actual situations. For example, the value range of the second number can be from 10 to 30. The specific value of the second number in the embodiments of the present invention is not limited.

[0071] Based on the spectral information and variety information of the third sample seeds, and the spectral information and variety information of the fourth sample seeds, a transfer matrix corresponding to the seeds to be identified is generated by means of spectral calibration transfer.

[0072] Specifically, based on the variety information of the third sample seeds and the variety information of the fourth sample seeds, an association relationship can be established between the third sample seeds and the fourth sample seeds with the same variety.

[0073] Based on the above association relationship, according to the spectral information of the third sample seeds with the variety being the target variety and the spectral information of the fourth sample seeds with the variety being the target variety , a transfer matrix corresponding to the target variety is generated by means of spectral calibration transfer.

[0074] By splicing the transfer matrices corresponding to each target variety, a transfer matrix corresponding to the seeds to be identified can be obtained .

[0075] As an optional embodiment, based on the spectral information and variety information of the third sample seeds, and the spectral information and variety information of the fourth sample seeds, a transfer matrix corresponding to the seeds to be identified is generated through spectral calibration transfer, including: based on the spectral information and variety information of the third sample seeds, and the spectral information and variety information of the fourth sample seeds, a transfer matrix corresponding to the seeds to be identified is generated through the direct standardization algorithm.

[0076] It should be noted that the direct standardization algorithm (DS) is an algorithm for data standardization or model transfer. The core idea of the direct standardization algorithm is to establish a conversion relationship or conversion matrix to minimize the differences between different data sources or instruments. The direct standardization algorithm usually involves establishing a certain functional relationship between the corresponding data of the standard set under two or more measurement conditions.

[0077] In the embodiment of the present invention, the spectral information of the third sample seeds with the target variety as the variety and the spectral information of the fourth sample seeds with the target variety as the variety are used to generate a transfer matrix corresponding to the target variety through the direct standardization algorithm.

[0078] The spectral information of the seeds to be identified is corrected based on the transfer matrix corresponding to the seeds to be identified, and the corrected spectral information of the seeds to be identified is obtained.

[0079] Specifically, after obtaining the transfer matrix corresponding to the seeds to be identified, the spectral information of the seeds to be identified can be corrected through numerical calculation based on the transfer matrix corresponding to the seeds to be identified, and the corrected spectral information of the seeds to be identified is obtained.

[0080] Optionally, in the embodiment of the present invention, the transfer matrix corresponding to the seeds to be identified can be calculated and the spectral information of the seeds to be identified X u are multiplied to obtain the corrected spectral information of the seeds to be identified X ut .

[0081] In an embodiment of the present invention, when it is determined that the number of varieties included in the target seeds is small, the first sample seeds with the variety being the target variety are determined as the second sample seeds. Then, a second number of seeds are randomly selected from the second sample seeds of each target variety as the third sample seeds. Furthermore, based on the spectral information and variety information of the third sample seeds, and the spectral information and variety information of the fourth sample seeds, a transfer matrix corresponding to the seeds to be identified is generated by means of spectral calibration transfer, so as to correct the spectral information of the seeds to be identified based on the transfer matrix corresponding to the seeds to be identified, and obtain the corrected spectral information of the seeds to be identified. This can more efficiently and accurately correct the influence of the heterogeneity of ecological factors on the spectral information of seeds, expand the applicable range of the seed variety recognition model, and enable the seed variety recognition model to identify more seed varieties under different ecological factors.

[0082] As an optional embodiment, based on the first quantity and the spectral information and variety information of the fourth sample seeds, correcting the spectral information of the seeds to be identified in the target seeds to obtain the corrected spectral information of the seeds to be identified includes: when it is determined that the first quantity is greater than the quantity threshold, clustering the spectral information of the target seeds, and dividing the target seeds into multiple seed groups according to the clustering result.

[0083] Specifically, if the first quantity is greater than the quantity threshold, it can indicate that the number of varieties included in the target seeds is small, and a second correction strategy can be selected to correct the spectral information of the seeds to be identified.

[0084] The specific content of the second correction strategy is as follows: clustering the spectral information of the target seeds, and dividing the target seeds into multiple seed groups according to the clustering result, where each seed group includes multiple target seeds.

[0085] In the embodiment of the present invention, can be used to represent the th seed group, , represents the number of seed groups, .

[0086] As an optional embodiment, clustering the spectral information of the target seeds includes: calculating the distance of the spectral information of the target seeds based on the K-means clustering algorithm to obtain the clustering result.

[0087] It should be noted that the K-means clustering algorithm is a commonly used unsupervised learning algorithm for dividing a data set into K different clusters. It finds the optimal cluster division through an iterative method, making the data points within the clusters as similar as possible, while the data points between the clusters are as different as possible.

[0088] For each subgroup, a transfer matrix corresponding to each subgroup is generated based on the spectral information and variety information of the fourth sample seeds in each subgroup.

[0089] Specifically, for the th seed group , based on the spectral information and variety information of the fourth sample seeds in the th seed group , the transfer matrix corresponding to the th seed group can be generated through numerical calculation, mathematical statistics, etc. F k .

[0090] As an optional embodiment, generating a transfer matrix corresponding to each subgroup based on the spectral information and variety information of the fourth sample seeds in each subgroup includes: based on the variety information of the fourth sample seeds in each subgroup, the first sample seeds with the same variety as the fourth sample seeds in each subgroup are determined as the fifth sample seeds corresponding to each subgroup.

[0091] Specifically, for the th seed group , based on the variety information of the fourth sample seeds in the th seed group , the first sample seeds with the same variety as the fourth sample seeds in the th seed group are determined as the fifth sample seeds corresponding to the th seed group .

[0092] Randomly select a third number of the fifth initial sample seeds corresponding to each subgroup as the sixth sample seeds corresponding to each subgroup.

[0093] Specifically, randomly select a third number of the fifth sample seeds corresponding to the th seed group as the sixth sample seeds corresponding to the th seed group .

[0094] It should be noted that the third number in the embodiments of the present invention can be determined based on prior knowledge and / or actual situations. For example, the value range of the third number can be between 10 and 30. The specific value of the third number in the embodiments of the present invention is not limited.

[0095] Based on the spectral information and variety information of the sixth sample seeds corresponding to each subgroup, as well as the spectral information and variety information of the fourth sample seeds in each subgroup, a transfer matrix corresponding to each subgroup is generated by means of spectral calibration transfer.

[0096] Specifically, based on the variety information of the sixth sample seeds corresponding to the th seed group and the variety information of the fourth sample seeds in the th seed group , the correlation relationship between the sixth sample seeds corresponding to the th seed group with the same variety and the fourth sample seeds in the th seed group can be established.

[0097] Based on the above correlation relationship, according to the spectral information of the sixth sample seeds corresponding to the th seed group with the same variety and the spectral information of the fourth sample seeds in the th seed group , a transfer matrix corresponding to the th seed group is generated by means of spectral correction transfer. F k .

[0098] Based on the transfer matrix corresponding to each subgroup, the spectral information of the seeds to be identified in each subgroup is modified to obtain the corrected spectral information of the seeds to be identified in each subgroup.

[0099] Specifically, after obtaining the transfer matrix corresponding to the th seed group F , the spectral information of the seeds to be identified in the k th seed group can be corrected by means of numerical calculation to obtain the corrected spectral information of the seeds to be identified in the th seed group . .

[0100] The corrected spectral information of the seeds to be identified in each subgroup is spliced to obtain the corrected spectral information of the seeds to be identified.

[0101] ​In an embodiment of the present invention, when it is determined that the number of varieties included in the target seeds is relatively large, the spectral information of the target seeds is clustered. According to the clustering result, the target seeds are divided into multiple seed groups. Based on the variety information of the fourth sample seeds in each seed group, the first sample seeds with the same variety as the fourth sample seeds in each seed group are determined as the fifth sample seeds corresponding to each seed group. Randomly select the third number of seeds from the fifth initial sample seeds corresponding to each seed group as the sixth sample seeds corresponding to each seed group. Based on the spectral information and variety information of the sixth sample seeds corresponding to each seed group, as well as the spectral information and variety information of the fourth sample seeds in each seed group, a transfer matrix corresponding to each seed group is generated by means of spectral calibration transfer. Then, based on the transfer matrix corresponding to each seed group, the spectral information of the seeds to be identified in each seed group is corrected. The spectral information of the seeds to be identified in each seed group after correction is spliced to obtain the spectral information of the seeds to be identified after correction. When the number of varieties included in the target seeds is relatively large, pre-clustering can improve the efficiency of correcting the spectral information of the seeds to be identified, can more efficiently and accurately correct the influence of the heterogeneity of ecological factors on the spectral information of the seeds, can expand the applicable range of the seed variety recognition model, and enable the seed variety recognition model to identify more seed varieties under different ecological factors.

[0102] To facilitate the understanding of the seed variety recognition method provided by the present invention, the following uses an example to illustrate the seed variety recognition method provided by the present invention.

[0103] The first sample seeds and the target seeds in this example both include 5 seeds of common wheat, 3 pairs of hybrid wheat seeds and their parents, a total of 13 varieties. The first sample seeds and the target seeds have different harvest years and different storage times. Table 1 is the detailed list of the first sample seeds and the target seeds. The varieties and ecological factors of the first sample seeds and the target seeds are shown in Table 1.

[0104] Table 1 Detailed list of the first sample seeds and the target seeds

[0105] As shown in Table 1, the varieties (i.e., each target variety) included in the target seeds are: Jingnong 16, Jingnong 17, Jingnong 18, Jingnong 1952, Jingnong 1974, Jingmai 9, Jingmai 11, Jingmai 18305Yhua68-1, 05Yhua68-2, CP730, BS1086, and BS237. The quantity of each target variety (i.e., the first quantity) is 13.

[0106] In this example, the spectral information of seeds in the 365nm - 970nm band can be collected using a VideometerLab4TM instrument (Videometer A / S, Denmark).

[0107] The number of the first sample seeds of each target variety is 100, and the total number of the first sample seeds is 1300.

[0108] Collect the spectral information of the first sample seeds X s After that, using the spectral information of the first sample seeds X s as the training samples and the variety information of the first sample seeds Y s as the sample labels, train the initial model constructed based on the linear discriminant analysis classifier model to obtain a trained seed variety recognition model.

[0109] After collecting the spectral information of the target seeds, select some target seeds for variety identification to obtain the fourth sample seeds of each target variety. Among them, the number of the fourth sample seeds of each target variety is 150, and the total number of the fourth sample seeds is 1950.

[0110] Select 80 seeds from the fourth sample seeds of each target variety as the verification seeds of each target variety to verify the effectiveness of the seed variety recognition method provided by the present invention. Then the number of the fourth sample seeds of each target variety is 70.

[0111] In the embodiments of the present invention, it can be used to represent the spectral information of the verification seeds.

[0112] Adopt the first correction strategy to correct the spectral information of the fourth sample seeds X l and the spectral information of the verification seeds to obtain the corrected spectral information of the fourth sample seeds X lt and the corrected spectral information of the verification seeds .

[0113] Specifically, in this example, 30 seeds can be randomly selected from the fourth sample seeds of each target variety, and 30 seeds can be randomly selected from the first sample seeds of each target variety as the third sample seeds of each target variety.

[0114] Based on the spectral information of the fourth sample seeds of each target variety randomly selected X l and the spectral information of the third sample seeds of each target variety X s, generate the transfer matrix corresponding to the verification seeds through the method of spectral calibration transfer , and then, based on the transfer matrix corresponding to the verification seeds , respectively correct the spectral information of the fourth sample seeds X l and the spectral information of the verification seeds to obtain the corrected spectral information of the fourth sample seeds X lt and the corrected spectral information of the verification seeds .

[0115] Respectively input the spectral information of the first sample seeds X s , the spectral information of the fourth sample seeds X l , the corrected spectral information of the fourth sample seeds X lt , the spectral information of the verification seeds and the corrected spectral information of the verification seeds into the above-mentioned seed variety identification model to obtain the spectral information of the first sample seeds X s , the spectral information of the fourth sample seeds X l , the corrected spectral information of the fourth sample seeds X lt , the spectral information of the verification seeds and the corrected spectral information of the verification seeds corresponding variety identification accuracy rates. Table 2 is the result analysis table of the first correction strategy.

[0116] Table 2 Result analysis table of the first correction strategy

[0117] As shown in Table 2, the seed variety identification model obtains relatively good classification results for the spectral information of the first sample seeds , and the identification accuracy rates are all above 99%.

[0118] However, when using the seed variety identification model to identify the variety of target seeds, whether it is the spectral information of the fourth sample seeds X l or the spectral information of the verification seeds , the identification accuracy rates of the seed variety identification model are all lower than 20%. The identification accuracy rate of the seed variety identification model has dropped significantly when identifying seeds with different ecological factors.

[0119] Adopt the first correction strategy for the spectral information of the fourth sample seeds Xl and the spectral information of the verification seeds After correction, the spectral information of the fourth sample seeds corrected by the seed variety identification model X lt and the spectral information of the verified seeds after correction The recognition accuracy has been greatly improved. However, after inputting the spectral information of the verification seeds into the seed variety identification model, the recognition accuracy of the target varieties corresponding to 5 common wheat seeds reaches 91.8%, but the recognition accuracy for all target varieties is only 64.5%.

[0120] Adopt the second correction strategy for the spectral information of the fourth sample seeds X l and the spectral information of the verification seeds to perform correction, and obtain the spectral information of the fourth sample seeds after correction and the spectral information of the verification seeds after correction .

[0121] Respectively input the spectral information of the first sample seeds X s , the spectral information of the fourth sample seeds X l , the spectral information of the fourth sample seeds after correction , the spectral information of the verification seeds the spectral information of the verification seeds after correction into the above-mentioned seed variety identification model to obtain the spectral information of the first sample seeds X s , the spectral information of the fourth sample seeds X l , the spectral information of the fourth sample seeds after correction , the spectral information of the verification seeds after correction corresponding variety recognition accuracy. Table 3 is the result analysis table of the second correction strategy.

[0122] Table 3 is the result analysis table of the second correction strategy

[0123] As shown in Table 3, the seed variety identification model obtains relatively good classification results for the spectral information of the first sample seeds , and the recognition accuracy is above 99%.

[0124] The variety recognition accuracy of the seed variety identification model for the spectral information of the verification seeds after correction is 85.1%, and compared with the first correction strategy, the variety recognition accuracy has increased by 64.8%.

[0125] Moreover, the spectral information of the fourth sample seed X l and the spectral information of the verification seed After being corrected by adopting the second correction strategy, the seed variety recognition model for the corrected spectral information of the fourth sample seed and the corrected spectral information of the verification seed The variety recognition accuracy rate has been significantly improved.

[0126] Therefore, the above examples verify that the seed variety recognition method provided by the present invention can improve the accuracy rate of recognizing the seed varieties with different ecological factors from the historical sample seeds without reducing the accuracy rate of recognizing the seed varieties with the same ecological factors as the historical sample seeds, and verify that when the number of each target variety is large, adopting the second correction strategy to correct the spectral information of the seeds to be recognized can more efficiently and accurately correct the influence of the heterogeneity of ecological factors on the spectral information of the seeds.

[0127] The seed variety recognition method provided by the present invention utilizes the corresponding relationship between the same varieties of target seeds and the first sample seeds in different batches, selects samples from the first sample seeds and the target seeds as the standard samples for calibration transfer. Through the calibration transfer mechanism, the spectral information characteristics of the first sample seeds are transferred to the spectral information of the seeds to be recognized in the target seeds, reducing the influence of the heterogeneity of ecological factors on the spectral information of the seeds to be recognized. For the case where there are many varieties included in the target seeds, a clustering algorithm is adopted to convert the target seeds into multiple small sample sets for calibration transfer to improve the effect of calibration transfer.

[0128] The seed variety recognition method provided by the present invention can also be applied to plant samples or animal samples for other variety recognition. At the same time, similar to powdery and horny seeds, frozen and normal seeds, such samples that can be classified and recognized by spectroscopy can also adopt the method provided by the present invention.

[0129] Figure 3 It is a schematic structural diagram of the seed variety recognition device provided by the present invention. The following combines Figure 3 Describe the seed variety recognition device provided by the present invention. The variety recognition device described below can be mutually corresponding and referred to with the seed variety recognition method provided by the present invention described above. As Figure 3 shown, the device includes: an information acquisition module 301, a spectral correction module 302, and a variety recognition module 303.

[0130] An information acquisition module 301 is configured to acquire the spectral information of target seeds and the variety information of the fourth sample seeds in the target seeds, determine the variety included in the target seeds as the target variety, acquire the quantity of each target variety as the first quantity, the varieties included in the fourth sample seeds are each target variety, and the target seeds have the same ecological factors, where the ecological factors include factors affecting seed cultivation and / or seed storage.

[0131] A spectral correction module 302 is configured to correct the spectral information of the seeds to be identified in the target seeds based on the first quantity, the spectral information and variety information of the fourth sample seeds, and acquire the corrected spectral information of the seeds to be identified, where the variety of the seeds to be identified is one of the target varieties.

[0132] A variety identification module 303 is configured to input the corrected spectral information of the seeds to be identified into a variety identification model, and acquire the variety information of the seeds to be identified output by the variety identification model.

[0133] Wherein, the variety identification model is obtained after being trained based on the spectral information and variety information of the first sample seeds; the varieties of the first sample seeds include each alternative variety corresponding to the seeds to be identified.

[0134] Specifically, the information acquisition module 301, the spectral correction module 302 and the variety identification module 303 are electrically connected.

[0135] In the seed variety identification device in the embodiment of the present invention, by correcting the spectral information of the seeds to be identified in the target seeds based on the quantity of the varieties included in the target seeds, the spectral information of the target seeds and the variety information of the fourth sample seeds in the target seeds, after obtaining the corrected spectral information of the seeds to be identified, inputting the corrected spectral information of the seeds to be identified into the seed variety identification model, and obtaining the variety information of the seeds to be identified output by the seed variety identification model, it can correct the spectral information of a large number of seeds to be identified with unknown varieties in the target seeds based on the spectral information of the fourth sample seeds with a small number of known varieties in the target seeds and the spectral information of the first sample seeds with known varieties used for training the seed variety identification model. By correcting the spectral information of the seeds to be identified, correct the influence of the heterogeneity of ecological factors on the spectral information of the seeds, so as to improve the accuracy of identifying the seed varieties with ecological factors different from those of the historical sample seeds without reducing the accuracy of identifying the seed varieties with the same ecological factors as the historical sample seeds, can improve the efficiency of seed variety identification, can provide a more accurate data basis for seed breeding, seed trade and agricultural production, can better ensure the safety and stability of agricultural production, improve the yield and quality of crops, promote the healthy development of the seed industry and meet the global market's demand for high-quality agricultural products.

[0136] Figure 4Illustrates a schematic diagram of the physical structure of an electronic device, as follows Figure 4 As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440. Among them, the processor 410, the communication interface 420, and the memory 430 complete communication with each other through the communication bus 440. The processor 410 can call the logical instructions in the memory 430 to execute the seed variety identification method, which includes: obtaining the spectral information of the target seed and the variety information of the fourth sample seed in the target seed, determining the variety included in the target seed as the target variety, obtaining the quantity of each target variety as the first quantity, the variety included in the fourth sample seed being each target variety, the target seeds having the same ecological factors, and the ecological factors including at least one of the seed storage time, the seed cultivation year, and the seed cultivation conditions; based on the first quantity and the spectral information and variety information of the fourth sample seed, correcting the spectral information of the seed to be identified in the target seed to obtain the corrected spectral information of the seed to be identified, the variety of the seed to be identified being one of each target variety; inputting the corrected spectral information of the seed to be identified into the variety identification model to obtain the variety information of the seed to be identified output by the variety identification model; wherein, the variety identification model is obtained by training based on the spectral information of the first sample seed and the variety information of the first sample seed; the variety of the first sample seed includes each alternative variety corresponding to the seed to be identified.

[0137] In addition, when the logical instructions in the above-mentioned memory 430 can be implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0138] On the other hand, the present invention also provides a computer program product, which includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the seed variety identification method provided by each of the above methods. The method includes: obtaining the spectral information of the target seeds and the variety information of the fourth sample seeds in the target seeds, determining the variety included in the target seeds as the target variety, obtaining the quantity of each target variety as the first quantity, the variety included in the fourth sample seeds being each target variety, the target seeds having the same ecological factors, where the ecological factors include at least one of the seed storage time, the seed cultivation year, and the seed cultivation conditions; based on the first quantity and the spectral information and variety information of the fourth sample seeds, correcting the spectral information of the seeds to be identified in the target seeds to obtain the corrected spectral information of the seeds to be identified, the variety of the seeds to be identified being one of each target variety; inputting the corrected spectral information of the seeds to be identified into the variety identification model to obtain the variety information of the seeds to be identified output by the variety identification model; where the variety identification model is obtained by training based on the spectral information and variety information of the first sample seeds; the variety of the first sample seeds includes each alternative variety corresponding to the seeds to be identified.

[0139] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it realizes the execution of the seed variety identification method provided by each of the above methods. The method includes: obtaining the spectral information of the target seeds and the variety information of the fourth sample seeds in the target seeds, determining the variety included in the target seeds as the target variety, obtaining the quantity of each target variety as the first quantity, the variety included in the fourth sample seeds being each target variety, the target seeds having the same ecological factors, where the ecological factors include at least one of the seed storage time, the seed cultivation year, and the seed cultivation conditions; based on the first quantity and the spectral information and variety information of the fourth sample seeds, correcting the spectral information of the seeds to be identified in the target seeds to obtain the corrected spectral information of the seeds to be identified, the variety of the seeds to be identified being one of each target variety; inputting the corrected spectral information of the seeds to be identified into the variety identification model to obtain the variety information of the seeds to be identified output by the variety identification model; where the variety identification model is obtained by training based on the spectral information and variety information of the first sample seeds; the variety of the first sample seeds includes each alternative variety corresponding to the seeds to be identified.

[0140] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative work.

[0141] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A seed variety identification method, characterized in that: include: Acquire spectral information of target seeds and variety information of a fourth sample seed among the target seeds, determine the varieties included in the target seeds as target varieties, acquire the quantity of each target variety as a first quantity, the varieties included in the fourth sample seeds are each target variety, and the target seeds have the same ecological factors, and the ecological factors include factors affecting seed cultivation and / or seed storage; Based on the spectral information and variety information of the first number and the fourth sample seeds, the spectral information of the seeds to be identified among the target seeds is corrected to obtain the corrected spectral information of the seeds to be identified, and the variety of the seeds to be identified is one of the target varieties; Inputting the corrected spectral information of the seed to be identified into a variety identification model, and obtaining the variety information of the seed to be identified output by the variety identification model; The variety recognition model is obtained after training based on the spectral information of the first sample seeds and the variety information of the first sample seeds; the varieties of the first sample seeds include the candidate varieties corresponding to the seeds to be identified.

2. The seed variety identification method according to claim 1, characterized in that: The step of correcting the spectral information of the to-be-identified seeds among the target seeds based on the spectral information and variety information of the first number and the fourth sample seeds to obtain the corrected spectral information of the to-be-identified seeds includes: In the case where it is determined that the first quantity is not greater than the quantity threshold, based on the variety information of the first sample seeds, determining the first sample seeds of the target variety as the second sample seeds; For each target variety, randomly selecting a second number of seeds from the second sample seeds of each target variety as third sample seeds; Based on the spectral information and variety information of the third sample seeds, and the spectral information and variety information of the fourth sample seeds, a transfer matrix corresponding to the seeds to be identified is generated by spectral calibration transfer; The spectral information of the seed to be identified is corrected based on the transfer matrix corresponding to the seed to be identified, so as to obtain the corrected spectral information of the seed to be identified.

3. The seed variety identification method according to claim 2, characterized in that: The step of correcting the spectral information of the to-be-identified seeds among the target seeds based on the spectral information and variety information of the first number and the fourth sample seeds to obtain the corrected spectral information of the to-be-identified seeds includes: In the case where it is determined that the first number is greater than a number threshold, clustering the spectral information of the target seeds, and dividing the target seeds into a plurality of seed groups according to the clustering result; For each seed group, based on the spectrum information and variety information of the fourth sample seed in each seed group, generating a transfer matrix corresponding to each seed group; Based on the transfer matrix corresponding to each seed group, modify the spectral information of the seeds to be identified in each seed group to obtain the modified spectral information of the seeds to be identified in each seed group; The corrected spectral information of the seeds to be identified in each of the seed groups is spliced ​​to obtain the corrected spectral information of the seeds to be identified.

4. The seed variety identification method according to claim 3, characterized in that: The step of generating a transfer matrix corresponding to each seed group based on the spectrum information and variety information of the fourth sample seed in each seed group includes: Based on the variety information of the fourth sample seed in each seed group, determining the first sample seed of the same variety as the fourth sample seed in each seed group as the fifth sample seed corresponding to each seed group; randomly selecting a third number of the fifth initial sample seeds corresponding to each seed group as the sixth sample seeds corresponding to each seed group; Based on the spectral information and variety information of the sixth sample seed corresponding to each seed group and the spectral information and variety information of the fourth sample seed in each seed group, a transfer matrix corresponding to each seed group is generated by spectral calibration transfer.

5. The seed variety identification method according to claim 2, characterized in that: The step of generating a transfer matrix corresponding to the seed to be identified by spectral calibration transfer based on the spectral information and variety information of the third sample seed and the spectral information and variety information of the fourth sample seed includes: Based on the spectral information and variety information of the third sample seeds and the spectral information and variety information of the fourth sample seeds, a transfer matrix corresponding to the seeds to be identified is generated through a direct standardization algorithm.

6. The seed variety identification method according to claim 3, characterized in that: The clustering of the spectral information of the target seeds includes: The spectral information of the target seeds is clustered based on the K-means clustering algorithm to obtain the clustering result.

7. A seed variety identification device, characterized in that: include: an information acquisition module, used for acquiring spectral information of target seeds and variety information of a fourth sample seed in the target seeds, determining the variety included in the target seeds as the target variety, acquiring the quantity of each target variety as the first quantity, the variety included in the fourth sample seeds is each target variety, and the target seeds have the same ecological factors, and the ecological factors include factors affecting seed cultivation and / or seed storage; a spectral correction module, used for correcting the spectral information of the seeds to be identified among the target seeds based on the spectral information and variety information of the first number and the fourth sample seeds, and obtaining the corrected spectral information of the seeds to be identified, wherein the variety of the seeds to be identified is one of the target varieties; A variety identification module, used for inputting the corrected spectral information of the seeds to be identified into a variety identification model, and obtaining the variety information of the seeds to be identified output by the variety identification model; The variety recognition model is obtained after training based on the spectral information of the first sample seeds and the variety information of the first sample seeds; the varieties of the first sample seeds include the candidate varieties corresponding to the seeds to be identified.

8. An electronic device 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 seed variety identification method according to any one of claims 1 to 6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the seed variety identification method according to any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the seed variety identification method according to any one of claims 1 to 6 is implemented.

Images