Fine management method and system for crop planting

Abstract

The invention provides a crop planting fine management method and system, and relates to the technical field of agricultural informationization, and the method comprises the steps: carrying out the itinerant shooting in a crop planting region, obtaining the sampling images of a plurality of crops in a plurality of subregions, and inputting a species recognition model, species information and growth stage information of crops are obtained; summarizing the sampling images to obtain a sampling image set of the crops with the same species information; determining whether the environmental parameters need to be adjusted or not according to the growth stage information and the environmental parameters of the crops in the sampling images in the sampling image set; if adjustment is needed, the adjusted environmental parameters are determined according to the preset growth stage, the growth stage information and the environmental parameters in the multiple partitions. According to the invention, whether the environmental parameters of each subarea need to be adjusted can be judged, thereby facilitating the management of growth stages and growth states of crops, and promoting the crops to reach an ideal growth stage.

Description

Technical Field

[0001] The present invention relates to the field of agricultural information technology, and in particular to a method and system for refined management of crop planting. Background Art

[0002] In related technologies, CN118014758A relates to the field of smart agricultural technology and discloses an intelligent management system for crop planting, including: an information data acquisition module, a network transmission unit, an information data storage unit, a core processing module, a planting plan adjustment unit, an abnormal alarm unit, a plan feedback unit, and a terminal display module; the information data acquisition module is used to collect information data on factors affecting planting; the network transmission unit is used to transmit the information data collected by the information data acquisition module over the network. By collecting weather information, soil information, and crop information, it is possible to analyze and determine the weather conditions, soil texture of the planting field, and the status of the crops over a period of time, and based on the analyzed and determined weather conditions and soil texture information of the planting field, it is possible to match the most suitable crops for planting, adjust the planting plan, and then feedback the adjusted planting plan to the management personnel.

[0003] CN119048264A relates to the field of agricultural information technology, particularly to a crop planting information collection and analysis method and system. The system polls crop seeds to obtain historical seed initiation data; uses a crop seed growth monitoring system and / or growth monitoring equipment to obtain crop seed growth monitoring data; obtains multiple stored germination experimental cycles from the crop seeds; wherein the germination experimental cycle is the difference between the storage timestamp and the experimental germination timestamp of each stored seed; and generates integrated crop seed vitality data based on the crop seed vitality data; wherein the vitality data includes historical seed initiation data, growth monitoring data, and germination experimental cycles. This solution can be used to select high-quality oilseed crop seeds, improve the germination rate of oilseed crops, and promote oilseed crop breeding improvements.

[0004] Therefore, relevant technologies can monitor the planting and growth conditions of crops through information technology, but it is difficult to manage and optimize the planting and growth environment of crops, and it is also difficult to manage the growth stage and growth status of crops, so it is difficult to make crops reach the ideal growth stage.

[0005] The information disclosed in the background technology section of this application is only intended to deepen the understanding of the general background technology of this application, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the Invention

[0006] The present invention provides a method and system for refined management of crop planting, which can solve the technical problem that related technologies are difficult to manage the growth stages and growth states of crops.

[0007] According to a first aspect of the present invention, there is provided a method for refined management of crop planting, comprising: Conducting roving photography in a crop planting area to obtain sampling images of multiple crops in multiple zones; Input multiple sampled images into the species recognition model to obtain the species type and growth stage information of the crops in each partition; Sampling images of the same species in different partitions are aggregated to obtain a set of crop sampling images with the same species information; determining whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the plurality of partitions; If adjustment is required, the adjusted environmental parameters are determined according to the preset growth stage, the growth stage information of the crops in each sampled image, and the environmental parameters in the multiple partitions.

[0008] According to a second aspect of the present invention, there is provided a crop planting refined management system, comprising: A shooting module is used to perform roving shooting in the crop planting area to obtain sampling images of multiple crops in multiple partitions; The recognition module is used to input multiple sample images into the species recognition model to obtain the species type information and growth stage information of the crops in each partition; A collection module is used to aggregate the sampling images of the same species in different partitions to obtain a collection of sampling images of crops with the same species information; a judgment module, configured to determine whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the multiple partitions; The adjustment module is used to determine the adjusted environmental parameters according to the preset growth stage, the growth stage information of the crops in each sampling image and the environmental parameters in multiple partitions if adjustment is required.

[0009] By adopting the above technical solution, the present invention can achieve the following technical effects: According to the present invention, a species recognition model can be used to identify the types and growth stages of crops planted in multiple zones. This allows for statistical analysis of the growth stage information of crops of the same type planted in different zones. Based on the statistical results, it can be used to determine whether the environmental parameters of each zone need to be adjusted to ensure that the crops in each zone are in an ideal growth environment. This facilitates the management of the crop's growth stage and growth status, and helps the crops reach their ideal growth stage. When training the species recognition model, a growth stage recognition module can be used to determine the crop's growth stage feature vector. The growth stage feature vector and the first sample feature map can then be used to comprehensively determine the crop's species and growth stage, thereby improving the accuracy of species recognition for crops at different growth stages. Furthermore, during training, the species recognition model can be trained through multiple training steps to improve its capabilities in various aspects, such as growth stage recognition and species type recognition, thereby improving training efficiency and the accuracy of the species recognition model. When determining the authenticity loss function, adversarial training can be used to determine the authenticity loss function, thereby improving the fidelity of images generated by the image generation model while also improving the discriminative ability of the discriminative model. This balances the performance of the two models and improves the fidelity of images generated by the first sample. When determining the generation accuracy loss function, the generation accuracy loss function can be formed by combining the cosine similarity between the generated species information and the labeled species information, and the cosine similarity between the generated growth stage information and the labeled growth stage information. This improves the consistency of the generated species information and the generated growth stage information with the labeled information of the first sample image during training, while reducing the error between the training species information and the training growth stage information and the labeled information of the first sample image, as well as the error between the generated species information and the training species information, and the error between the generated growth stage information and the training growth stage information. This improves both the accuracy of the image generation model and the accuracy of the species recognition model in identifying the species and growth stage of crops in the image. When determining the second comprehensive loss function, the second branch can be used to automatically supervise the training of the first branch, i.e., the second branch can participate in the accuracy competition of the first branch, thereby urging the first branch to improve its accuracy during training. Furthermore, the second branch can also continuously improve its accuracy during training, thereby participating in the competition at a higher standard, i.e., supervising the training of the first branch at a higher standard, thereby improving the accuracy of the first branch, i.e., improving the accuracy of the species recognition model and improving training efficiency.When determining the third comprehensive loss function, the appearance errors of the crops in the first sample image and the first sample generated image in terms of morphology and color can be used to determine the third comprehensive loss function, and the first training probability distribution can be used as an amplification coefficient in the solution process to reflect the influence of the authenticity of the image generated by the image generation model on the second morphological feature information and the second color feature information, so as to act in the training of the image generation model and the species recognition model to amplify the appearance error. At the same time, the first training probability distribution can also be improved to further improve the authenticity of the image generated by the image generation model and improve the training efficiency.

[0010] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and not limiting of the present invention. Other features and aspects of the present invention will become more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. Those skilled in the art can derive other embodiments based on these drawings without inventive efforts. Figure 1 A schematic diagram exemplarily illustrates a process flow of a method for refined crop planting management according to an embodiment of the present invention; Figure 2 A schematic diagram exemplarily illustrates the training of a species recognition model according to an embodiment of the present invention; Figure 3 A block diagram of a crop planting refined management system according to an embodiment of the present invention is exemplarily shown. DETAILED DESCRIPTION

[0012] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0013] The following specific embodiments are used to describe the technical solution of the present invention in detail. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0014] Figure 1A flow chart of a method for refined crop planting management according to an embodiment of the present invention is exemplarily shown. The method includes: Step S101, performing roving photography in a crop planting area to obtain sampling images of a plurality of crops in a plurality of partitions; Step S102: inputting the plurality of sampled images into a species recognition model to obtain species information and growth stage information of the crops in each partition; Step S103, summarizing the sampled images of the same species in different partitions to obtain a set of sampled images of crops with the same species information; Step S104, determining whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the multiple partitions; Step S105 : If adjustment is required, the adjusted environmental parameters are determined according to the preset growth stage, the growth stage information of the crops in each sampled image, and the environmental parameters in the multiple partitions.

[0015] According to the refined management method of crop planting in an embodiment of the present invention, the types and growth stages of crops planted in multiple partitions can be identified through a species recognition model, so that the growth stage information of the same type of crops planted in different partitions can be statistically analyzed, and then based on the statistical results, it can be judged whether the environmental parameters of each partition need to be adjusted, so that the crops in each partition are in an ideal growth environment, which is conducive to the management of the growth stage and growth status of the crops, and promotes the crops to reach the ideal growth stage.

[0016] According to one embodiment of the present invention, in step S101, the planting area may be a vegetable greenhouse or other planting area. The vegetable greenhouse may include multiple partitions, each of which is relatively independent and can independently adjust environmental parameters, such as air temperature, air humidity, soil moisture, light intensity, and daily light duration. Patrol photography may be performed manually or using equipment such as drones and robots to obtain sample images of crops within each partition. Each partition may only be planted with one crop, and when acquiring sample images, sample images of multiple crops may be obtained for each partition.

[0017] According to one embodiment of the present invention, in step S102, the species recognition model can be used to obtain species information and growth stage information of crops in the sampling image. Both the species information and the growth stage information can be information in vector form. The species recognition model can obtain the species information and growth stage information of multiple sampling images in a unified partition, and solve the average value to obtain the species information and growth stage information of the crops in the partition.

[0018] Figure 2 A schematic diagram exemplarily illustrates the training of a species recognition model according to an embodiment of the present invention.

[0019] According to one embodiment of the present invention, the species recognition model is a deep learning neural network model, for example, a convolutional neural network model. The training steps of the species recognition model include: inputting a first sample image into an encoding module of the species recognition model to obtain a first sample feature map of the first sample image, wherein the first sample image is an image of a crop of a specific species at a specific growth stage; inputting the first sample feature map into a growth stage recognition module of the species recognition model to obtain a growth stage feature vector of the crop in the first sample image; inputting the growth stage feature vector and the first sample feature map into a decoding module of the species recognition model to obtain training species type information and training growth stage information of the crop in the first sample image; obtaining a first sample generation image based on the training species type information and the training growth stage information, and an image generation model; obtaining a first comprehensive loss function based on the first sample generation image, a discriminant model, and a species recognition model; determining a second comprehensive loss function based on the growth stage feature vector, the training species type information, and the training growth stage information; obtaining a third comprehensive loss function based on the first sample generation image, the first sample image, and an image feature recognition model; training the species recognition model based on the first comprehensive loss function, the second comprehensive loss function, and the third comprehensive loss function to obtain a trained species recognition model.

[0020] According to one embodiment of the present invention, during the training process, a first sample image can be used for training. The first sample image is an image of a crop of a specific species at a specific growth stage, and the first sample image has labeled information, including labeled species information and labeled growth stage information, which can serve as error-free species information and growth stage information of the crop in the first sample image.

[0021] According to one embodiment of the present invention, the encoding module of the species recognition model may include multiple layers such as convolutional layers, activation layers, and pooling layers, and may perform encoding processing on the first sample image to obtain a first sample feature map of the first sample image. The number of first sample feature maps is more than the first sample image, but the resolution of each first sample feature map is lower than the first sample image, and each first sample feature map can be used to represent the characteristics of one aspect of the first sample image.

[0022] According to one embodiment of the present invention, the growth stage identification module may also include multiple layers, such as convolution layers, activation layers, pooling layers, etc., which can be used to process the first sample feature map. Multiple first sample feature maps can be input into the growth stage identification module respectively to obtain a growth stage feature vector. The growth stage feature vector is vector type data and can be used to describe the growth stage of the crop, such as the germination stage, seedling stage, tillering stage, jointing stage, ear growth stage, and fruiting stage.

[0023] According to one embodiment of the present invention, the growth stage feature vector and the first sample feature map can be input into the decoding module of the species recognition model to obtain the training species type information and training growth stage information of the crops in the first sample image. The decoding module may include multiple layers such as deconvolution layers and fully connected layers, which can improve the resolution of the first sample feature map, reduce the number of first sample feature maps, and finally output the training species type information and training growth stage information through the fully connected layer. The training species type information can be used to describe the species of the crops in the first sample image. The training species type information is similar to the above-mentioned species type information and can be data in the form of a vector. The training growth stage information is similar to the above-mentioned growth stage information and can also be data in the form of a vector and can be used to determine the growth stage of the crops. For example, each component of the vector is the probability of multiple periods. For example, the probability of being in the germination period is 1%, the probability of being in the seedling period is 1%, the probability of being in the tillering period is 1%, the probability of being in the jointing period is 1%, the probability of being in the earing period is 1%, and the probability of being in the fruiting period is 95%. The training growth stage information is .

[0024] According to one embodiment of the present invention, the image generation model is also a deep learning neural network model, which can be used to generate images based on information represented by vectors. For example, based on the training species type information and the training growth stage information, the corresponding image of the crop can be generated, that is, a first sample generation image is generated. Furthermore, based on the characteristics of the first sample generation image, the error of the training species type information and the training growth stage information output by the species recognition model can be determined.

[0025] According to one embodiment of the present invention, a first comprehensive loss function is obtained based on the first sample generated image, the discriminant model and the species recognition model, including: inputting the first sample generated image into the discriminant model to obtain a first training probability distribution that the first sample generated image is a real image; inputting the first sample image into the discriminant model to obtain a second training probability distribution that the first sample image is a real image; inputting the first sample generated image into the species recognition model to obtain generated species type information and generated growth stage information of the first sample generated image; determining the first comprehensive loss function based on the first training probability distribution, the second training probability distribution, the generated species type information, the generated growth stage information and the annotation information of the first sample image.

[0026] According to one embodiment of the present invention, the discriminant model is a deep learning neural network model, such as a convolutional neural network model, which can be used to determine whether an input image is a real-world image or a model-generated virtual image, and can output a probability that the image is a real-world image. A first sample generated image can be input into the discriminant model to obtain a first training probability distribution, and the first sample image can be input into the discriminant model to obtain a second training probability distribution. Furthermore, the first sample generated image can be input into a species recognition model to obtain information about the generated species and the generated growth stage of the first sample generated image. During the training process, the authenticity of the images generated by the image generation model can be continuously improved through training, and the discrimination accuracy of the discriminant model can be continuously improved, so that the performance of the two can be improved together. In this way, when the discrimination accuracy of the discriminant model is high, the image generation model can still generate images with high enough authenticity, which is enough to make the discriminant model difficult to distinguish. Moreover, during training, the species types of crops in the images generated by the image generation model can be made to conform to the description of the training species type information, and the growth stage of crops in the images generated by the image generation model can be made to conform to the description of the training growth stage information, so that the generated species type information and generated growth stage information obtained by the first sample generated image input species recognition model are consistent with the labeling information of the first sample image. Since the first sample generated image is generated through the image generation model The training species type information and the training growth stage information are obtained based on the first sample image, and the generated species type information and the generated growth stage information are obtained by the species recognition model based on the first sample image. Therefore, by making the generated species type information and the generated growth stage information consistent with the annotation information of the first sample image, the species recognition model and the image generation model can be mutually verified and trained simultaneously, that is, the error between the training species type information and the training growth stage information and the annotation information of the first sample image, the error between the generated species type information and the training species type information, and the error between the generated growth stage information and the training growth stage information are reduced at the same time, thereby simultaneously improving the accuracy of the image generated by the image generation model and the accuracy of the species recognition model in identifying the species types and growth stages of crops in the image.

[0027] According to one embodiment of the present invention, a first comprehensive loss function is determined based on the first training probability distribution, the second training probability distribution, the generated species type information, the generated growth stage information and the annotation information of the first sample image, including: obtaining the annotated species type information and the annotated growth stage information of the crops in the first sample image based on the annotation information of the first sample image; determining the authenticity loss function based on the first training probability distribution and the second training probability distribution; determining the generation accuracy loss function based on the generated species type information, the generated growth stage information, the annotated species type information and the annotated growth stage information; and determining the first comprehensive loss function based on the authenticity loss function and the generation accuracy loss function.

[0028] According to one embodiment of the present invention, determining the authenticity loss function according to the first training probability distribution and the second training probability distribution includes: determining the authenticity loss function according to formula (1) , (1) in, Generate the first training probability distribution of the image as the real image for the i-th first sample, is the second training probability distribution that the i-th first sample image is a true image, n is the number of first sample images in a training batch, i≤n, and i and n are both positive integers, Indicates that The direction of minimization adjusts the parameters of the image generation model, Indicates that The direction of maximization adjusts the parameters of the discriminant model.

[0029] According to one embodiment of the present invention, theoretically, the probability that the discriminant model judges the first sample image as a real image is 100%, and the probability that the first sample generated image is judged as a real image is 0. However, as the training process proceeds, the realism of the generation model becomes higher and higher, resulting in an increase in the probability that the discriminant model judges the first sample generated image as a real image. Moreover, as the discriminant model also improves its discrimination ability during training, the probability that the discriminant model judges the first sample generated image as a real image may again decrease. Ultimately, the performance of the discriminant model and the image generation model can be balanced, so that even if the discrimination accuracy of the discriminant model is very high, it is still difficult to judge whether the first sample generated image generated by the image generation model is a real image, that is, the realism of the image generated by the image generation model is made higher.

[0030] According to one embodiment of the present invention, according to the above-mentioned training method for balancing the performance of the image generation model and the discriminant model, the discriminant model can be adjusted in the direction of maximizing the authenticity loss function expressed by formula (1), and the image generation model can be adjusted in the direction of maximizing the authenticity loss function. In formula (1), if the input image is the first sample image, then in the training of the discriminant model, Maximize, that is, as close to 1 as possible, so that Maximize, thereby improving the accuracy of the discriminant model in judging the real image. If the input image is the first sample generated image, then in the training of the discriminant model, Minimize, that is, as close to 0 as possible, so that Maximize, thereby improving the accuracy of the discriminant model in judging the generated images.

[0031] According to one embodiment of the present invention, on the other hand, in formula (1), if the input image is the first sample generated image, then in the training of the image generation model, Maximize, that is, as close to 1 as possible, so that Minimizing the probability that the first sample generated image generated by the image generation model is recognized as a real image by the discrimination model is increased, thereby improving the realism of the image generated by the image generation model and improving the performance of the image generation model.

[0032] In this way, the authenticity loss function can be determined through adversarial training, which can improve the realism of the images generated by the image generation model while improving the discriminative ability of the discriminative model, thereby balancing the performance of the two models and improving the realism of the images generated by the first sample.

[0033] According to one embodiment of the present invention, determining a generation accuracy loss function based on the generated species information, the generated growth stage information, the annotated species information, and the annotated growth stage information includes: determining a generation accuracy loss function according to formula (2): , (2) in, Generate species information for the i-th first sample image, Generate the growth stage information of the image for the i-th first sample, is the labeled species information of the i-th first sample image, is the label growth stage information of the i-th first sample image, for The transposed vector of for The transposed vector of , n is the number of the first sample images in a training batch, i≤n, and i and n are both positive integers, and The preset weights.

[0034] According to one embodiment of the present invention, The cosine similarity between the generated species information and the annotated species information can be used to describe the consistency between the generated species information and the annotated species information. The cosine similarity between the generated growth stage information and the labeled growth stage information can be used to describe the consistency between the generated growth stage information and the labeled growth stage information. When the generated species information and the generated growth stage information are consistent with the labeled information of the first sample image, the above two cosine similarities are theoretically 1. Therefore, the error between 1 and the above two cosine similarities can constitute a generation accuracy loss function, that is, the error between 1 and the above two cosine similarities is weighted and summed, and the above weighted sum value corresponding to each first sample image in a training batch is summed to obtain the generation accuracy loss function, and the generation accuracy loss function is reduced during the training process, thereby reducing the error between 1 and the above two cosine similarities and improving the consistency between the generated species information and the generated growth stage information and the labeled information of the first sample image.

[0035] In this way, an accuracy loss function can be generated by forming the cosine similarity between the generated species information and the labeled species information, as well as the cosine similarity between the generated growth stage information and the labeled growth stage information. Therefore, during the training process, the consistency between the generated species information and the generated growth stage information and the labeled information of the first sample image is improved, while the error between the training species information and the training growth stage information and the labeled information of the first sample image, the error between the generated species information and the training species information, and the error between the generated growth stage information and the training growth stage information are reduced, thereby simultaneously improving the accuracy of the image generation model in generating images and the species recognition model in recognizing the species and growth stages of crops in the image.

[0036] According to one embodiment of the present invention, the above-mentioned generation accuracy loss function and authenticity loss function may be weightedly summed to obtain a first comprehensive loss function.

[0037] According to one embodiment of the present invention, emphasis may be placed on improving the accuracy of the growth stage feature vector, thereby improving the accuracy of the training species type information and the training growth stage information obtained based on the growth stage feature vector.

[0038] According to one embodiment of the present invention, a second comprehensive loss function is determined based on the growth stage feature vector, the training species type information and the training growth stage information, including: decoding the growth stage feature vector through a growth stage information decoding module to obtain first growth stage information; obtaining the labeled species type information and labeled growth stage information of the crops in the first sample image based on the labeling information of the first sample image; and determining the second comprehensive loss function based on the first growth stage information, the training species type information, the training growth stage information, the labeled species type information and the labeled growth stage information.

[0039] According to one embodiment of the present invention, the growth stage information decoding module may include layers such as a deconvolution layer and a fully connected layer, and may decode the growth stage feature vector separately to obtain the first growth stage information.

[0040] According to one embodiment of the present invention, determining a second comprehensive loss function based on the first growth stage information, the training species type information, the training growth stage information, the labeled species type information, and the labeled growth stage information includes: determining the second comprehensive loss function according to formula (3) , (3) in, is the training growth stage information of the i-th first sample image, is the label growth stage information of the i-th first sample image, for The transposed vector of is the first growth stage information corresponding to the i-th first sample image, for The transposed vector of is the training species information of the i-th first sample image, for The transposed vector of is the labeled species information of the i-th first sample image, n is the number of first sample images in a training batch, i≤n, and i and n are both positive integers, 、 and is the preset weight, and if is the conditional function.

[0041] According to one embodiment of the present invention, the conditional function Indicates that the conditions are met In the case of , otherwise, the conditional function value is . Through this conditional function, the two branches can be trained in a competitive manner, wherein the first branch is a branch composed of an encoding module, a growth stage identification module and a decoding module, that is, a branch that obtains training species information and training growth stage information, and the second branch is a branch composed of an encoding module, a growth stage identification module and a growth stage information decoding module, that is, a branch that obtains the first growth stage information. When the above conditions of the conditional function are met, the cosine similarity between the training growth stage information and the labeled growth stage information is greater than or equal to the cosine similarity between the first growth stage information and the labeled growth stage information, that is, the accuracy of the first branch is higher. In this case, represents the error between the training growth stage information and the labeled growth stage information, Represents the error between the training species information and the labeled species information, Represents the error between the first growth stage information and the labeled growth stage information, and the three errors are weighted and summed to obtain the conditional function value, which is used as the second comprehensive loss function corresponding to the i-th first sample image, so that during the training process, the second comprehensive loss function is minimized, thereby reducing the above three errors. The above two branches can be trained at the same time, and the training intensity of the two branches is determined by the preset weights of the three errors. If the above conditions of the conditional function are not met, that is, the accuracy of the second branch is higher, then use To amplify the error between the training growth stage information and the labeled growth stage information, as well as the error between the training species information and the labeled species information, the weighted sum of the amplified errors is used as the conditional function value. In this case, It is only used as an amplification factor. Therefore, during the training process, only the error between the training growth stage information and the labeled growth stage information and the error between the training species information and the labeled species information are involved in the training. Therefore, only the first branch is trained with a larger training intensity to improve the accuracy of the first branch. And, use As an amplification factor, it can be used to reflect the impact of inaccurate eigenvectors in the growth stage on the error of the species identification model. Therefore, when training using the above method, when the accuracy of the first branch is lower than that of the second branch, the amplification factor can be used to increase the training intensity of the first branch, thereby improving the training efficiency of the first branch. When the accuracy of the first branch is higher than or equal to that of the second branch, the first branch and the second branch are trained at the same time to simultaneously improve the accuracy of the first branch and the second branch. In other words, the second branch is used to automatically supervise the training of the first branch, or the second branch is used to participate in the accuracy competition of the first branch, thereby urging the first branch to improve its accuracy during training. Moreover, the second branch also continuously improves its accuracy during training, thereby participating in the competition with higher standards, or supervising the training of the first branch with higher standards, thereby improving the accuracy of the species identification model.

[0042] In this way, the second branch can be used to automatically supervise the training of the first branch, that is, the second branch is used to participate in the accuracy competition of the first branch, thereby urging the first branch to improve its accuracy during training. Moreover, the second branch also continuously improves its accuracy during training, thereby participating in the competition with a higher standard, that is, supervising the training of the first branch with a higher standard, thereby improving the accuracy of the first branch, that is, improving the accuracy of the species recognition model and improving training efficiency.

[0043] According to one embodiment of the present invention, the species recognition model and the image generation model can be trained based on the similarity between the first sample generated image and the first sample image. Since the first sample generated image is generated by the image generation model based on the training species type information and the training growth stage information, and the training species type information and the training growth stage information are obtained by the species recognition model based on the first sample image, when the accuracy of the species recognition model and the image generation model is high, the consistency of the species type and growth stage of the crops contained in the first sample image and the first sample generated image is high. If there is an error in the species type and growth stage of the two, there is an error in the appearance of the crops contained in the two, so that a third comprehensive loss function can be obtained based on the error in the appearance, and the species recognition model and the image generation model can be further trained by the third comprehensive loss function to improve the accuracy of the species recognition model and the image generation model.

[0044] According to one embodiment of the present invention, a third comprehensive loss function is obtained based on the first sample generated image and the first sample image, as well as an image feature recognition model, including: inputting the first sample image into the image feature recognition model to obtain first morphological feature information and first color feature information of crops in the first sample image; inputting the first sample generated image into the image feature recognition model to obtain second morphological feature information and second color feature information of crops in the first sample generated image; and determining the third comprehensive loss function based on the first morphological feature information, the first color feature information, the second morphological feature information, the second color feature information and the first training probability distribution.

[0045] According to one embodiment of the present invention, an image feature recognition model can be used to identify morphological feature information and color feature information (both in vector form) of a target contained in an input image. For example, the morphology of multiple parts (e.g., leaves, flowers, stems, etc.) of a crop in an image can correspond to one or more components of the morphological feature information, and the colors of multiple parts of the crop in the image can correspond to one or more components of the color feature information. Therefore, the image feature recognition model can obtain first morphological feature information and first color feature information of a first sample image, and can also obtain second morphological feature information and second color feature information of an image generated by the first sample. The image feature recognition model is a deep learning neural network model, such as a convolutional neural network model. The present invention does not limit the specific type of the image feature recognition model.

[0046] According to one embodiment of the present invention, determining a third comprehensive loss function based on the first morphological feature information, the first color feature information, the second morphological feature information, the second color feature information and the first training probability distribution includes: determining the third comprehensive loss function according to formula (4) , (4) in, is the first morphological feature information of the i-th first sample image, Generate the second morphological feature information of the image for the i-th first sample, is the first color feature information of the i-th first sample image, Generate the second color feature information of the image for the i-th first sample, Generate the first training probability distribution of the image as the real image for the i-th first sample, and is the preset weight, n is the number of first sample images in a training batch, i≤n, and both i and n are positive integers.

[0047] According to one embodiment of the present invention, in formula (4), is the error between the first morphological feature information and the second morphological feature information, The error between the first color feature information and the second color feature information can be weighted and summed up, thereby reducing the result of the weighted summation during training to reduce the two errors and improve the accuracy of the species recognition model and the image generation model. Furthermore, the first training probability distribution can be used as an amplification factor to improve training efficiency. Using the first training probability distribution as an amplification factor can be used to reflect the influence of the authenticity of the image generated by the image generation model on the second morphological feature information and the second color feature information, so that it can be used in the training of the image generation model and the species recognition model to amplify the above-mentioned error and improve training efficiency. At the same time, it can also improve the first training probability distribution and further improve the authenticity of the image generated by the image generation model.

[0048] In this way, the appearance errors of crops in morphology and color in the first sample image and the first sample generated image can be used to determine the third comprehensive loss function, and the first training probability distribution can be used as an amplification coefficient in the solution process to reflect the influence of the authenticity of the image generated by the image generation model on the second morphological feature information and the second color feature information, so as to act on the training of the image generation model and the species recognition model to amplify the appearance error, and at the same time, the first training probability distribution can also be improved, thereby further improving the authenticity of the image generated by the image generation model and improving the training efficiency.

[0049] According to one embodiment of the present invention, the species recognition model can be trained based on the above-mentioned first comprehensive loss function, second comprehensive loss function and third comprehensive loss function, that is, the species recognition model is trained through the above training method, and in the process of training through each loss function, the parameters of the species recognition model can be adjusted when the loss function is back-propagated, so that the training of the species recognition model can be assisted by a variety of other models (image generation model, discrimination model, image feature recognition model, growth stage information decoding module), thereby improving the training efficiency and training effect of the species recognition model. After multiple trainings, a trained species recognition model can be obtained, which can be used for fine-grained identification of crop species and growth stages.

[0050] According to an embodiment of the present invention, in step S103, sampling images of the same species in different partitions may be aggregated to obtain a set of sampling images of crops with the same species information, and the partitions corresponding to these sampling images may be labeled.

[0051] According to one embodiment of the present invention, in step S104, it can be determined whether the environmental parameters within each partition need to be adjusted. Determining whether the environmental parameters within each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters within the multiple partitions includes: obtaining preset growth stage vector information corresponding to the preset growth stage; determining the growth stage similarity between the growth stage information of the crops in each sampled image and the preset growth stage vector information; setting an adjustment judgment condition for the growth stage similarity; and determining whether the environmental parameters within each partition need to be adjusted if the growth stage similarities corresponding to the each sampled image meet the adjustment judgment condition.

[0052] According to one embodiment of the present invention, the preset growth stage is an expected growth stage estimated based on the duration of crop planting, and the preset growth stage vector information is a growth stage vector corresponding to the expected growth stage. The format of the vector is similar to the above-mentioned training growth stage information and will not be repeated here.

[0053] According to one embodiment of the present invention, if the environmental parameters are suitable, then the type of crops can theoretically grow to the above-mentioned preset growth stage within the above-mentioned planting time. However, the environmental parameters will affect the growth of the crops. Therefore, the crops in each partition may have different growth stages due to different environments. The growth stage information determined by the above-mentioned species recognition model can be compared with the preset growth stage vector information. If the growth stage similarity between the two (for example, cosine similarity) is low, the environmental parameters in the partition are not suitable for crop growth and the environmental parameters should be adjusted.

[0054] According to one embodiment of the present invention, setting adjustment judgment conditions for growth stage similarity includes: determining the average value and standard deviation of the growth stage similarity, as well as the maximum value and minimum value of the growth stage similarity; determining the adjustment judgment conditions to be at least one of the following: the average value of the growth stage similarity is less than or equal to the average value threshold; the standard deviation of the growth stage similarity is greater than or equal to the standard deviation threshold; the maximum value of the growth stage similarity is less than or equal to the maximum value threshold; the minimum value of the growth stage similarity is less than or equal to the minimum value threshold.

[0055] According to one embodiment of the present invention, if the average value of the growth stage similarity is less than or equal to the average value threshold, it means that the average level of the suitability of the environmental parameters of each partition is not up to standard, resulting in unsatisfactory growth conditions of the crops in each partition. If the standard deviation of the growth stage similarity is greater than or equal to the standard deviation threshold, it means that the environments of each partition are quite different, resulting in large differences in the growth conditions of the crops in each partition. If the maximum value of the growth stage similarity is less than or equal to the maximum value threshold, it means that the environmental parameters of each partition are not suitable. If the minimum value of the growth stage similarity is less than or equal to the minimum value threshold, it means that there are partitions with unsuitable environmental parameters. Therefore, if any one of the above adjustment judgment conditions is met, it can be determined whether the environmental parameters in each partition need to be adjusted. When adjusting, all of them may need to be adjusted, or only the environmental parameters of some partitions may need to be adjusted.

[0056] According to one embodiment of the present invention, in step S105, if adjustment is required, the adjusted environmental parameters are determined based on the preset growth stage, the growth stage information of the crops in each sampling image, the species type information, and the environmental parameters in multiple partitions, including: inputting the growth stage information and species type information of the crops in the sampling image, as well as the planting time of the crops in each partition during the current patrol shooting into the environmental parameter estimation model to obtain the predicted environmental parameters of each partition; determining the loss function of the environmental parameter estimation model based on the environmental parameters of each partition and the predicted environmental parameters; training the environmental parameter estimation model based on the loss function of the environmental parameter estimation model to obtain the trained environmental parameter estimation model; obtaining the preset growth stage vector information corresponding to the preset growth stage; inputting the planting time of the crops in each partition during the next patrol shooting and the preset growth stage vector information into the trained environmental parameter estimation model to determine the adjusted environmental parameters.

[0057] According to one embodiment of the present invention, the environmental parameter estimation model is a deep learning neural network model, such as a BP neural network model. Suitable environmental parameters can be determined using the environmental parameter estimation model. The environmental parameter estimation model can be trained using information about the growth stages of the crops in each of the aforementioned zones and the planting duration. The environmental parameter estimation model can then derive predicted environmental parameters for each zone. Specifically, the environmental parameter estimation model can estimate under what environmental parameters a particular type of crop will grow to a particular state within the aforementioned planting duration.

[0058] According to one embodiment of the present invention, the environmental parameters output by the environmental parameter estimation model may have errors, which can be compared with the environmental parameters actually detected in each partition to obtain the loss function of the environmental parameter estimation model. The environmental parameter estimation model is then trained through the loss function of the environmental parameter estimation model to reduce the errors between the predicted environmental parameters output by the environmental parameter estimation model and the environmental parameters actually detected in each partition, thereby improving the accuracy of the predicted environmental parameters. The weighted sum of the relative errors between multiple predicted environmental parameters and multiple environmental parameters can be used as the loss function. After backpropagation of the loss function determined by the relative errors between multiple predicted environmental parameters and multiple environmental parameters in multiple partitions, the trained environmental parameter estimation model is obtained.

[0059] According to one embodiment of the present invention, the preset growth stage vector information and the planting time of the crops in each partition during the next patrol shooting can be input into the trained environmental parameter estimation model to determine the adjusted environmental parameters, and the environmental parameters of the partitions whose growth stage similarity is less than or equal to the similarity threshold are adjusted, that is, set to the adjusted environmental parameters so that during the next patrol shooting, the growth stage of the crops in these partitions reaches or is close to the preset growth stage.

[0060] Figure 3 A block diagram of a crop planting refined management system according to an embodiment of the present invention is exemplarily shown, and the system includes: A shooting module is used to perform roving shooting in the crop planting area to obtain sampling images of multiple crops in multiple partitions; The recognition module is used to input multiple sample images into the species recognition model to obtain the species type information and growth stage information of the crops in each partition; A collection module is used to aggregate the sampling images of the same species in different partitions to obtain a collection of sampling images of crops with the same species information; a judgment module, configured to determine whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the multiple partitions; The adjustment module is used to determine the adjusted environmental parameters according to the preset growth stage, the growth stage information of the crops in each sampling image, the species information and the environmental parameters in multiple partitions if adjustment is required.

[0061] According to one embodiment of the present invention, determining whether environmental parameters in each partition need to be adjusted based on the growth stage information of the crop in each sampled image in the sampled image set and the environmental parameters in the plurality of partitions includes: Obtaining preset growth stage vector information corresponding to the preset growth stage; Determining the growth stage similarity between the growth stage information of the crops in each sampled image and the preset growth stage vector information; Set the adjustment judgment conditions for the similarity of growth stages; When the similarities of the growth stages corresponding to the respective sampled images meet the adjustment judgment conditions, it is determined whether the environmental parameters within the respective partitions need to be adjusted.

[0062] According to one embodiment of the present invention, setting the adjustment judgment condition of the growth stage similarity includes: Determine the mean and standard deviation of the similarity between growth stages, as well as the maximum and minimum values of the similarity between growth stages; The adjustment judgment condition is determined to be at least one of the following: The average value of the similarity of the growth stages is less than or equal to the average threshold; The standard deviation of the similarity of the growth stages is greater than or equal to the standard deviation threshold; The maximum value of similarity in the growth stage is less than or equal to the maximum threshold; The minimum value of the similarity in the growth stage is less than or equal to the minimum threshold.

[0063] According to one embodiment of the present invention, if adjustment is required, the adjusted environmental parameters are determined based on the preset growth stage, the growth stage information of the crops in each sampled image, the species information, and the environmental parameters in the multiple partitions, including: The growth stage and species information of the crops in the sampled images, as well as the planting duration of the crops in each partition during this tour, are input into the environmental parameter estimation model to obtain the predicted environmental parameters of each partition; Determining a loss function of an environmental parameter estimation model based on the environmental parameters of each partition and the predicted environmental parameters; The environmental parameter estimation model is trained according to the loss function of the environmental parameter estimation model to obtain a trained environmental parameter estimation model; Obtaining preset growth stage vector information corresponding to the preset growth stage; The planting time of the crops in each partition during the next patrol shooting and the preset growth stage vector information are input into the trained environmental parameter estimation model to determine the adjusted environmental parameters.

[0064] According to the method for refined management of crop planting according to an embodiment of the present invention, the species and growth stages of crops planted in multiple partitions can be identified through a species recognition model, so that the growth stage information of the same type of crops planted in different partitions can be statistically analyzed, and then based on the statistical results, it is determined whether the environmental parameters of each partition need to be adjusted, so that the crops in each partition are in an ideal growth environment, which is conducive to the management of the growth stage and growth state of the crops, and promotes the crops to reach an ideal growth stage. When training the species recognition model, the growth stage feature vector of the crop can be determined through the growth stage recognition module, and then the species type and growth stage of the crop can be comprehensively determined through the growth stage feature vector and the first sample feature map, thereby improving the species recognition accuracy of crops at different growth stages. Moreover, during the training process, the species recognition model can be trained in multiple aspects such as growth stage recognition and species type recognition through multiple training processes, thereby improving the training efficiency and the accuracy of the species recognition model. When determining the authenticity loss function, the authenticity loss function can be determined through adversarial training, thereby improving the realism of the images generated by the image generation model while improving the discriminative ability of the discriminative model, thereby balancing the performance of the two models and improving the realism of the images generated by the first sample. When determining the generation accuracy loss function, the generation accuracy loss function can be composed of the cosine similarity between the generated species information and the labeled species information, and the cosine similarity between the generated growth stage information and the labeled growth stage information, so that during the training process, the consistency of the generated species information and the generated growth stage information with the labeled information of the first sample image is improved, while reducing the error between the training species information and the training growth stage information and the labeled information of the first sample image, the error between the generated species information and the training species information, and the error between the generated growth stage information and the training growth stage information, thereby simultaneously improving the accuracy of the images generated by the image generation model and the accuracy of the species recognition model in identifying the species and growth stages of crops in the image. When determining the second comprehensive loss function, the second branch can be used to automatically supervise the training of the first branch, that is, the second branch is used to participate in the accuracy competition of the first branch, thereby urging the first branch to improve its accuracy during training. Moreover, the second branch also continuously improves its accuracy during training, thereby participating in the competition with a higher standard, that is, supervising the training of the first branch with a higher standard, thereby improving the accuracy of the first branch, that is, improving the accuracy of the species recognition model and improving training efficiency.When determining the third comprehensive loss function, the appearance errors of the crops in the first sample image and the first sample generated image in terms of morphology and color can be used to determine the third comprehensive loss function, and the first training probability distribution can be used as an amplification coefficient in the solution process to reflect the influence of the authenticity of the image generated by the image generation model on the second morphological feature information and the second color feature information, so as to act in the training of the image generation model and the species recognition model to amplify the appearance error. At the same time, the first training probability distribution can also be improved to further improve the authenticity of the image generated by the image generation model and improve the training efficiency.

[0065] Those skilled in the art will appreciate that the embodiments of the present invention described above and shown in the accompanying drawings are intended to be illustrative only and are not intended to limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and illustrated in the embodiments. Any variations or modifications may be made to the embodiments of the present invention without departing from the principles described.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for refined management of crop planting, characterized in that: include: Conducting roving photography in a crop planting area to obtain sampling images of multiple crops in multiple zones; Input multiple sampled images into the species recognition model to obtain the species type and growth stage information of the crops in each partition; Sampling images of the same species in different partitions are aggregated to obtain a set of crop sampling images with the same species information; determining whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the plurality of partitions; If adjustment is required, the adjusted environmental parameters are determined based on the preset growth stage, the growth stage information of the crops in each sampling image, the species information, and the environmental parameters in the multiple partitions.

2. The crop planting refined management method according to claim 1, characterized in that: Determining whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the plurality of partitions includes: Obtaining preset growth stage vector information corresponding to the preset growth stage; Determining the growth stage similarity between the growth stage information of the crops in each sampled image and the preset growth stage vector information; Set the adjustment judgment conditions for the similarity of growth stages; When the similarities of the growth stages corresponding to the respective sampled images meet the adjustment judgment conditions, it is determined whether the environmental parameters within the respective partitions need to be adjusted.

3. The crop planting refined management method according to claim 2, characterized in that: Set the adjustment judgment conditions for the similarity of the growth stage, including: Determine the mean and standard deviation of the similarity between growth stages, as well as the maximum and minimum values of the similarity between growth stages; The adjustment judgment condition is determined to be at least one of the following: The average value of the similarity of the growth stages is less than or equal to the average threshold; The standard deviation of the similarity of the growth stages is greater than or equal to the standard deviation threshold; The maximum value of similarity in the growth stage is less than or equal to the maximum threshold; The minimum value of the similarity in the growth stage is less than or equal to the minimum threshold.

4. The crop planting refined management method according to claim 1, characterized in that: If adjustment is required, the adjusted environmental parameters are determined based on the preset growth stage, the growth stage information of the crops in each sampled image, the species information, and the environmental parameters in multiple partitions, including: The growth stage and species information of the crops in the sampled images, as well as the planting duration of the crops in each partition during this tour, are input into the environmental parameter estimation model to obtain the predicted environmental parameters of each partition; Determining a loss function of an environmental parameter estimation model based on the environmental parameters of each partition and the predicted environmental parameters; The environmental parameter estimation model is trained according to the loss function of the environmental parameter estimation model to obtain a trained environmental parameter estimation model; Obtaining preset growth stage vector information corresponding to the preset growth stage; The planting time of the crops in each partition during the next patrol shooting and the preset growth stage vector information are input into the trained environmental parameter estimation model to determine the adjusted environmental parameters.

5. A crop planting refined management system, characterized in that: include: A shooting module is used to perform roving shooting in the crop planting area to obtain sampling images of multiple crops in multiple partitions; The recognition module is used to input multiple sample images into the species recognition model to obtain the species type information and growth stage information of the crops in each partition; A collection module is used to aggregate the sampling images of the same species in different partitions to obtain a collection of sampling images of crops with the same species information; a judgment module, configured to determine whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the multiple partitions; The adjustment module is used to determine the adjusted environmental parameters according to the preset growth stage, the growth stage information of the crops in each sampling image, the species information and the environmental parameters in multiple partitions if adjustment is required.

6. The crop planting refined management system according to claim 5, characterized in that: Determining whether the environmental parameters in each partition need to be adjusted based on the growth stage information of the crops in each sampled image in the sampled image set and the environmental parameters in the plurality of partitions includes: Obtaining preset growth stage vector information corresponding to the preset growth stage; Determining the growth stage similarity between the growth stage information of the crops in each sampled image and the preset growth stage vector information; Set the adjustment judgment conditions for the similarity of growth stages; When the similarities of the growth stages corresponding to the respective sampled images meet the adjustment judgment conditions, it is determined whether the environmental parameters within the respective partitions need to be adjusted.

7. The crop planting refined management system according to claim 6, characterized in that: Set the adjustment judgment conditions for the similarity of the growth stage, including: Determine the mean and standard deviation of the similarity between growth stages, as well as the maximum and minimum values of the similarity between growth stages; The adjustment judgment condition is determined to be at least one of the following: The average value of the similarity of the growth stages is less than or equal to the average threshold; The standard deviation of the similarity of the growth stages is greater than or equal to the standard deviation threshold; The maximum value of similarity in the growth stage is less than or equal to the maximum threshold; The minimum value of the similarity in the growth stage is less than or equal to the minimum threshold.

8. The crop planting refined management system according to claim 5, characterized in that: If adjustment is required, the adjusted environmental parameters are determined based on the preset growth stage, the growth stage information of the crops in each sampled image, the species information, and the environmental parameters in multiple partitions, including: The growth stage and species information of the crops in the sampled images, as well as the planting duration of the crops in each partition during this tour, are input into the environmental parameter estimation model to obtain the predicted environmental parameters of each partition; Determining a loss function of an environmental parameter estimation model based on the environmental parameters of each partition and the predicted environmental parameters; The environmental parameter estimation model is trained according to the loss function of the environmental parameter estimation model to obtain a trained environmental parameter estimation model; Obtaining preset growth stage vector information corresponding to the preset growth stage; The planting time of the crops in each partition during the next patrol shooting and the preset growth stage vector information are input into the trained environmental parameter estimation model to determine the adjusted environmental parameters.

Images