Method and system for predicting insect pests of dispersedly planted crops based on multimode data

Abstract

The embodiment of the invention relates to the technical field of information, in particular to a decentralized planting crop pest prediction method and system based on multimode data. The method comprises the following steps: reading a map of a target area, obtaining a water body position and a terrain type according to the map, and reading temperature data and rainfall data of the target area within a preset duration; reading an aerial image of the target area within a preset duration, and identifying crop types, plant populations and animal individuals in the target area; predicting an animal population according to the plant population, and correcting the predicted animal population by using the animal individuals; according to the water body position, the terrain type, the crop type, the temperature data and the rainfall data, predicting the pest type and the pest scale; and correcting the pest scale according to the animal population, and obtaining a crop pest prediction result according to the corrected pest scale and the crop type.

Description

TECHNICAL FIELD

[0001] Embodiments of the present specification relate to the field of information technology, and in particular to a dispersed planting crop pest prediction method and system based on multi-modal data. BACKGROUND

[0002] The dispersed planting mode still belongs to the main crop planting mode in some areas. Due to the wide distribution of dispersed planting areas and relatively extensive management, the monitoring and control of crop pests face great challenges. Traditional pest prediction methods mainly rely on manual investigation and statistical analysis of historical weather data, which have the disadvantages of poor timeliness, limited coverage, strong subjectivity, etc., and are difficult to meet the needs. With the development of unmanned aerial vehicle technology, geographic information system (GIS) and meteorological data, crop pest prediction has obtained new technical means. Some studies attempt to use satellite remote sensing images to obtain vegetation index to assess crop health status, or combine meteorological factors to establish a simple pest occurrence model. However, these methods mostly only focus on a single type of data, and fail to fully integrate multi-source heterogeneous information such as geographical environment, meteorological conditions, and biological community dynamics, resulting in low prediction accuracy, especially in complex terrain and small-scale dispersed planting areas. Therefore, there is an urgent need for a high-precision crop pest prediction method suitable for dispersed planting scenarios. SUMMARY

[0003] Embodiments of the present specification describe a dispersed planting crop pest prediction method and system based on multi-modal data.

[0004] In a first aspect, the embodiments of the present specification provide a dispersed planting crop pest prediction method based on multi-modal data, comprising the steps of:

[0005] reading a map of a target area, obtaining water body positions and terrain types according to the map, reading temperature data and precipitation data in a preset time period of the target area;

[0006] reading aerial images of the target area in the preset time period, identifying crop types, plant populations and animal individuals in the target area;

[0007] predicting an animal population according to the plant population, and correcting the predicted animal population using the animal individuals;

[0008] predicting pest types and pest scales according to the water body positions, the terrain types, the crop types, the temperature data and the precipitation data;

[0009] correcting the pest scale according to the animal population, and obtaining a prediction result of crop pest according to the corrected pest scale and the crop type.

[0010] In a second aspect, the embodiments of the present specification provide a dispersed planting crop pest prediction system based on multi-mode data, comprising:

[0011] a reading module configured to read a map of a target area, obtain water body positions and terrain types according to the map, and read temperature data and precipitation data within a preset time period of the target area;

[0012] a recognition module configured to read aerial images of the target area within the preset time period, and identify crop types, plant populations and animal individuals in the target area;

[0013] a correction module configured to predict animal populations according to the plant populations, and correct the predicted animal populations using the animal individuals;

[0014] a prediction module configured to predict pest types and pest scales according to the water body positions, the terrain types, the crop types, the temperature data and the precipitation data;

[0015] a result module configured to correct the pest scales according to the animal populations, and obtain a prediction result of crop pests according to the corrected pest scales and the crop types.

[0016] In a third aspect, the embodiments of the present specification provide an electronic device, comprising a processor and a memory;

[0017] the processor is connected to the memory;

[0018] the memory is configured to store executable program codes;

[0019] the processor runs a program corresponding to the executable program codes by reading the executable program codes stored in the memory, so as to execute the method of any one of the above aspects.

[0020] In a fourth aspect, the embodiments of the present specification provide a computer readable storage medium having a computer program stored thereon, wherein the computer program is executed by a processor to implement the method of any one of the above aspects.

[0021] In a fifth aspect, the embodiments of the present specification provide a computer program product comprising a computer program, wherein the computer program is executed by a processor to implement the method of any one of the above aspects.

[0022] The technical solutions provided by some embodiments of the present specification have at least the following beneficial effects:

[0023] In the embodiments of the present specification, the multi-mode data-based scattered planting crop pest prediction method and system provided by the present specification introduces animal-plant ecological correlation information by fusing geographic information, meteorological data, aerial images, and ecological correlation models, predicts animal population based on plant population distribution, and dynamically corrects using actually observed animal individual data, combines pest suitable breeding environment information and food chain balance model, so that the prediction result is closer to the real ecological system operation mechanism. The off-axis angle correction and space-time alignment of multi-time period aerial images improve the accuracy of crop type, plant population and animal individual identification. By establishing the pest tolerance limit of crop type, automatic alarm of pest risk is realized, which provides data support for pest control and helps to overcome the problems of single data, neglect of ecological relationship and large prediction deviation of traditional methods.

[0024] Other features and advantages of the embodiments of the present specification will be further disclosed in the following specific embodiments and drawings. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly explain the technical solutions in the embodiments of the present specification, the drawings required to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present specification, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0026] Figure 1 The scattered planting crop pest prediction schematic diagram provided by the present specification.

[0027] Figure 2 The scattered planting crop pest prediction method flowchart provided by the present specification.

[0028] Figure 3 The aerial image recognition schematic diagram provided by the present specification.

[0029] Figure 4 The scattered planting crop pest prediction system schematic diagram provided by the present specification.

[0030] Figure 5 The electronic device schematic diagram provided by the present specification. DETAILED DESCRIPTION

[0031] The technical solutions of the embodiments of the present specification will be explained and described below in combination with the drawings of the embodiments of the present specification. However, the following embodiments are only preferred embodiments of the present specification, not all. Based on the embodiments in the embodiments, other embodiments obtained by those skilled in the art without creative labor also belong to the protection scope of the present specification.

[0032] The terms "first", "second", "third", and the like in the description and claims of the present specification and the above drawings are used to distinguish different objects, and are not used to describe a particular order. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device including a series of steps or units is not limited to the listed steps or units, but can optionally further include steps or units not listed, or can optionally further include other steps or units inherent to the process, method, product, or device.

[0033] In the following description, the indications of orientation or positional relationship such as the terms "inner", "outer", "upper", "lower", "left", "right", and the like are only for the convenience of describing the embodiments and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present specification.

[0034] The data involved in the present application are all information and data authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data comply with relevant laws, regulations, and standards of relevant countries and regions.

[0035] Before introducing the technical solutions recorded in the present specification, the application scenarios of the technical solutions and related technologies are introduced.

[0036] The application of aerial photography technology in the field of agriculture, especially in the monitoring of crop areas, has important significance and wide-ranging effects. By using unmanned aerial vehicles, manned aircraft, or near-ground aerial platforms equipped with optical, multispectral, hyperspectral, and thermal infrared sensors, aerial photography of farmland can quickly obtain large-scale, high-resolution crop growth information, providing key data support for precision agriculture management. Compared with traditional manual patrols, aerial photography can quickly cover large areas of farmland, significantly improving monitoring efficiency, and is particularly suitable for scattered planting and complex terrain areas. High-resolution images can clearly present information such as crop growth, planting boundaries, missing plants, abnormal water and fertilizer, etc. Through image recognition and spectral analysis, early identification of crop characteristics (such as leaf discoloration, curling, and spots) infected by pests or diseases can be achieved, providing a reference for prevention and control. Multi-temporal aerial photography can construct a crop growth time series graph, enabling continuous monitoring of the growth cycle and phenological changes for yield estimation. In addition to crops, aerial photography can also capture information such as field weeds, insect activity traces, and individual animals such as birds and predatory insects, providing a basis for ecological system balance analysis.

[0037] The UAV flight height is low, and the shooting is easy to produce off-axis distortion. In addition, the crop canopy may cause the lower layer of pests to be difficult to observe. The existing application focuses on the crop itself, and less information such as animal individuals 23 and plant communities observed in aerial photography is included in the ecological chain model, which limits more accurate prediction of pest occurrence mechanism.

[0038] For some crop types 24, there is no obvious morphological feature in the seedling stage, which makes it difficult to accurately distinguish by relying only on high-definition visible light images. The cost of hyperspectral imaging is relatively high, and its application in scattered planting areas is limited.

[0039] In the hilly areas of the south, agricultural production is mainly small-scale and scattered planting, and farmland is scattered in the valleys between the mountains, gentle slope areas and around water bodies. The crops planted include rice, corn, tea and various economic forest fruits. Due to the complex terrain and poor transportation, the traditional pest patrol method is inefficient, and it is often difficult to find pests in time, resulting in delayed prevention and control and causing yield reduction. At the same time, the climate in this region is humid, with frequent rainfall during the rainy season and suitable temperature, which is easy to breed pests such as rice planthoppers, borers, leafhoppers and other pests. Combined with the large fluctuations in the population of natural enemy animals in the ecosystem, such as spiders, frogs and insectivorous birds, pest occurrence is sudden and hidden.

[0040] The present specification provides a technology for crop type 24 identification and pest prediction using aerial images 12 collected by relatively low-altitude flight. Please refer to the accompanying drawings Figure 1 , the present specification constructs a digital farmland file covering the region by periodically obtaining high-resolution UAV aerial images 12 in the region, combining GIS geographic information platform, real-time temperature and humidity and precipitation data from weather stations, and historical land use layers. After automatic correction and identification of the aerial images 12, the crop types 24 are divided, the plant population 25 density is counted, and the animal individuals 23 and their spatio-temporal distribution characteristics in the field are captured. Further, by integrating the ecological knowledge base, the dynamic relationship between plants, pests and natural enemies is analyzed, and the pest outbreak risk is predicted.

[0041] For example, after a continuous rainfall process, according to the information such as the increase of soil moisture, the rise of air temperature and the expansion of rice planting area, it is predicted that the rice planthopper breeding conditions are mature, and a moderate pest warning is initially generated. At the same time, through aerial identification, it is found that the number of insectivorous birds in the ridge area is reduced by 30% compared with the same period last year, and the prediction result is corrected by combining the food chain model, and finally the warning level is upgraded to "high risk", and is located to multiple key prevention and control areas. Agricultural technicians deploy insect traps, release natural enemy insects or carry out targeted pesticide application in advance to contain the spread of pests.

[0042] Specifically, the present specification first provides a scattered planting crop pest prediction method based on multi-modal data, please refer to the accompanying drawings Figure 2 , comprising the steps of:

[0043] Step S1) reading a map 11 of the target area, obtaining water body locations 21 and terrain types 22 from the map 11, reading temperature data and precipitation data in a preset time period of the target area. The map 11 is a GIS map 11, which contains rich spatial layer information. Through the analysis of the map 11, the location and distribution range of the water bodies (such as rivers, lakes, ponds, irrigation channels, etc.) in the region can be identified and extracted. Water bodies are not only an important factor affecting the local microclimate (such as increasing air humidity, adjusting temperature), but also a potential breeding ground for many pests (such as mosquitoes, aquatic insects and their larvae), so their spatial distribution is of great significance to the prediction of the occurrence of specific pests. According to the comprehensive analysis of the land use / cover data (Land Use / Land Cover, LULC) and the digital elevation model (Digital Elevation Model, DEM) in the map 11, the terrain type 22 information is obtained. The land use map is used to identify different land surface types such as vegetation cover area, bare land, farmland, forest land, etc.; the digital elevation model provides elevation, slope, slope direction and other terrain features. Through the fusion analysis of the two, the terrain categories such as mountains, hills, plains and depressions can be classified. Different terrain categories are suitable for different pests to inhabit and breed. For example, low-lying land is suitable for the breeding of rice planthoppers. The low-lying land is low and flat, with poor drainage, easy to accumulate water, and the soil is mostly clay, with strong water retention, often found along the river or in the valley between the mountains. Rice planthoppers are one of the main pests of rice, and their nymphs and adults prefer high humidity and shaded environments, and are suitable for breeding in rice fields with high humidity and poor ventilation. The low-lying land has frequent water accumulation, the air humidity is maintained above 80% for a long time, and the rice is planted densely, providing an ideal habitat and breeding site for rice planthoppers.

[0044] Reading the meteorological observation data of the target area in a preset time period (such as the past 6 months, 1 year or 2 years), mainly including temperature and precipitation data. It can be sourced from regional weather stations, remote sensing inversion products or weather data service platforms. Temperature data reflects environmental heat conditions, directly affecting the development rate, reproduction period and overwintering survival rate of pests. Precipitation data affects soil moisture, plant surface wetting time and water body area change, which is a key indicator for determining whether pests, especially moisture-loving pests, have breeding conditions.

[0045] Reading the map 11 of the target area, obtaining the water body locations 21 and terrain types 22 from the map 11 includes:

[0046] Reading the land use and cover map, obtaining the locations and ranges of water bodies and vegetation from the land use and cover map;

[0047] Reading the digital elevation model, obtaining the locations and ranges of mountains from the digital elevation model;

[0048] According to the location and range of the mountains, water bodies, and vegetation, the terrain type 22 of the target area is obtained.

[0049] A land use and cover map is read, and the land surface is divided into different functional categories, such as water bodies, farmland, forest land, grassland, and construction land, through remote sensing image interpretation or field survey marking. According to the identification of the "water body" category in the map, the geographical location, boundary range, and area information of water body elements such as rivers, lakes, reservoirs, ponds, and ditches are extracted. At the same time, the distribution of vegetation cover areas (such as forest land, grassland, and farmland) is identified. A digital elevation model (DEM) is read. DEM is a three-dimensional data model that stores elevation values in a regular grid form. Through the calculation and analysis of terrain factors such as slope, slope direction, relief, and elevation variability of DEM data, the distribution location and spatial range of mountains are identified. For example, by setting an elevation threshold and a slope gradient, mountainous, hilly, and plain areas can be distinguished. Combined with terrain curvature analysis, mountain peaks, slopes, and footprints are further identified.

[0050] After obtaining the spatial distribution information of water bodies, vegetation, and mountains, comprehensive discrimination is performed to generate the terrain type 22 classification results of the target area. Through spatial analysis algorithms, the water body distribution is combined with elevation characteristics to identify landform units such as river valleys, depressions, and tidal flats. The vegetation cover is combined with terrain undulation to determine whether the forest land is located on a steep slope or in a valley. According to the pre-set terrain classification rules, the target area is divided into mountainous, hilly, plain, basin, river valley, and waterfront, which are terrain types 22 with ecological significance.

[0051] For example, according to the land use and cover map, the water bodies in the map are identified to include a natural river running from north to south, several scattered breeding ponds, and an irrigation canal system running through farmland, from which the boundaries of all water body elements are extracted to obtain their location and distribution. At the same time, forest land and tea plantations are identified for subsequent vegetation cover analysis.

[0052] Then read the digital elevation model data, by calculating the elevation value of each grid unit and the height difference of adjacent units, analyze the terrain features. And set the elevation of 200 meters above, slope greater than 15 ° area for mountain, identify a piece of continuous highland in the western region as mountain. Further through the slope classification and topographic profile analysis, distinguish the mountain top, steep slope and gentle slope area. Subsequently, the water distribution layer and the mountain distribution layer are spatially superimposed, and the vegetation cover information is comprehensively judged. Finally, the 50-meter-wide strip along the river, the elevation below 100 meters, and the slope less than 5 ° are determined as the river valley plain. The area located in the south of the mountain, with an elevation of 120-200 meters, a slope of 8°-15°, and covered with tea garden, is classified as hilly slope land. The low-lying water area between the river and the mountain is identified as the waterfront lowland because of the long-term existence of small ponds and wet vegetation. The western high-elevation steep slope area is marked as mountain forest area.

[0053] Step S2) read the aerial image 12 of the target area within a preset time period, identify the crop type 24, plant population 25 and animal individual 23 in the target area.

[0054] By analyzing the aerial images 12 of the target area obtained within a preset time period, the key biological elements are identified to provide dynamic biological data support for pest prediction. Read multiple aerial images 12 of the area within a certain time period, which are usually taken by visible light cameras carried by drones or aerial platforms, with high spatial resolution. And this embodiment uses low-altitude photography, although the distortion of the area with large edge off-axis angle is relatively larger, but this embodiment identifies and deletes the distortion. The image taken by low-altitude photography has higher clarity and is more suitable for the identification of animal individuals 23 and pests or other insects. Image recognition technology is used to process the aerial images 12, automatically detect and classify the ground targets in the image. Through the deep learning model, the system identifies different types of crops (such as rice, corn, wheat, etc.), divides their planting areas and boundaries, and forms a crop type 24 distribution map. Detect the animal individuals 23 in the image, such as birds, insects, frogs, etc. which may be related to pests, record their species, location and appearance time. These information together constitute the dynamic portrait of the biological community of the target area, which provides an important basis for subsequent analysis of the ecological relationship between plants and animals, prediction of the distribution of natural enemies of pests and correction of the size of pests.

[0055] Specifically, it includes:

[0056] Calculate the off-axis angle of each pixel in the aerial image 12, delete the aerial image 12 area with off-axis angle greater than a preset threshold, and the remaining aerial image 12 area is subjected to image recognition, and the aerial image 12 is divided into image blocks according to the identified objects;

[0057] The shooting position and camera parameter information of the aerial image 12 are obtained, the image block is aligned with the map 11 of the target area, and the shooting time, position mark, and shooting angle mark of the image block are obtained;

[0058] According to the objects, position marks, and shooting angle marks identified from all image blocks, the crop types 24 and regions in the target area are obtained;

[0059] According to the objects and position marks identified from all image blocks, the repeated plant populations 25 are aggregated, and the plant populations 25 in the target area are obtained;

[0060] According to the objects, shooting time, position marks, and shooting angle marks identified from all image blocks, the animal individuals 23 in the target area are obtained, and the positions and times at which the animal individuals 23 appear are marked.

[0061] The off-axis angle of each pixel in the aerial image 12 relative to the camera optical axis is calculated. Since the edge region of the image is prone to distortion and resolution decline due to the tilt of the viewing angle, the region with an off-axis angle greater than a preset threshold is removed, and the image region with smaller geometric distortion in the central viewing angle range is retained to improve the accuracy of subsequent identification. Please refer to FIG. 2 for an example. Figure 3 For example, if the off-axis angle of a certain pixel point exceeds a set threshold (such as 30°), it is considered that there is a large geometric distortion in this region, which is not suitable for accurate analysis, and is removed.

[0062] The retained image region is subjected to semantic segmentation or target detection to identify objects such as crops, vegetation, and animals, and the image is divided into multiple image blocks containing specific ground object categories according to the identification results, each image block corresponding to an identified object and its spatial range. The shooting position and camera parameters (such as focal length, pixel size, installation attitude, etc.) of the aerial image 12 are obtained, and the photogrammetry principle is used to georeference each image block with the GIS map 11 of the target area to achieve spatial alignment. In this process, each image block is assigned accurate geographic position coordinates, shooting time, and shooting angle marks to ensure that it can be fused with other data in a unified spatio-temporal coordinate system.

[0063] Finally, through spatial superposition analysis, the distribution regions of different crop types 24 are determined, and the crop type 24 distribution is generated. For plant populations 25, the individuals or regions of the same plant that repeatedly appear in different image blocks are marked and aggregated according to the position to identify continuous or aggregated plant community structures. For animal individuals 23, the position and time information of the animal individuals 23 is recorded in combination with their identification results, position, shooting time, and angle marks. Bird and insect animal individuals 23 that are active around farmland or tea gardens are identified, and their positions and times of appearance are recorded. This can be used to analyze which animals may be potential beneficial insects (help control the number of pests) or natural enemies of pests.

[0064] Step S3) Correcting the predicted animal population using the animal individual 23 according to the plant population 25.

[0065] The method for predicting an animal population according to a plant population 25, and correcting the predicted animal population using the animal individual 23 comprises:

[0066] Receiving information about the ecological relationship between animals and plants, which records the habitat support capacity of the plant population 25 to the animal population;

[0067] Based on the distribution of the plant population 25 in the target area, an initial animal population prediction result is generated by reasoning from the information about the ecological relationship between animals and plants, which includes the type, spatial distribution probability and estimated number of the animal population;

[0068] According to the animal population prediction result, animal habits and the plant population 25, the estimated number distribution of the animal individual 23 taken at each time period is obtained;

[0069] The identified animal individual 23, the location and time of the appearance of the animal individual 23 are compared with the estimated number distribution to obtain a correction coefficient;

[0070] The predicted animal population is corrected using the correction coefficient.

[0071] Data about the ecological relationship between animals and plants, i.e. information about the ecological relationship between animals and plants, records which animal population different plant populations 25 can support and their habitat capacity, i.e. how much suitable living environment or food resources certain plant species can provide for certain animals. Based on the distribution of the plant population 25 in the target area and the above information about the ecological relationship between animals and plants, the system can infer the types of animal populations that may exist in the area, their probability of appearing in each area and their estimated number. For example, if there is a large amount of vegetation suitable for insects to inhabit in a certain area, it can be inferred that there may be more insect species that depend on these vegetation for survival in the area. Combined with known animal habits and plant population 25 information, the number distribution of animal individuals 23 expected to appear in each time period is calculated. This step takes into account the behavior patterns of animals (such as foraging time, activity range, etc.) to more accurately estimate the spatio-temporal distribution of animals.

[0072] The actual animal individuals 23, the positions and times of appearance identified directly from the aerial image 12 by image recognition technology are compared with the estimated quantity distribution calculated previously. The degree of difference between the actual observation value and the predicted value is determined, and a correction proportion coefficient is calculated. The initial predicted animal population size is adjusted using the correction proportion coefficient obtained. If it is found that the actual number of a certain animal is significantly lower than expected, the estimate of the size of the population of that animal needs to be reduced.

[0073] In a certain rice planting area, first, the rice plants and weed communities on the ridges in the area are identified by aerial image 12. According to the preset animal-plant ecological correlation information, it is obtained that the rice field provides a foraging environment for sparrows, and the observation data of the similar area in the past few years are obtained, and the average density of sparrows is about 5 per hectare. The weeds on the ridges provide habitat and nesting conditions for the white-headed babbler, and the average capacity is obtained by obtaining the observation data of the similar area in the past few years, which is about 2 per 100 meters of ridge. Based on this, combined with the spatial distribution of rice and weeds, according to the average value obtained by observation, the initial animal population prediction is generated, and the total number of sparrows in the entire area is about 80, mainly distributed above the rice field, and the white-headed babbler is about 30, concentrated along the ridge line, and the appearance probability distribution of each area is given. Combined with the activity habits of birds (such as sparrows being active in the early morning and late afternoon) and the distribution of vegetation, it is predicted that about 60% of sparrows and 40% of white-headed babbler should be observed in the aerial photography time period, that is, about 48 sparrows and 12 white-headed babbler are estimated to be photographed, forming the estimated quantity distribution of each period.

[0074] The actual aerial recognition result shows that a total of 30 sparrows and 10 white-headed babbler are detected in the same area and period. By comparing the actual observation value with the estimated value, the correction proportion coefficient of the sparrow is calculated as 30 / 48=0.625, and the correction proportion coefficient of the white-headed babbler is calculated as 10 / 12≈0.83. The initial prediction is adjusted using the correction coefficient: the total number of sparrows is corrected from 80 to 80x0.625=50, and the total number of white-headed babbler is corrected from 30 to 30x0.83≈25. The corrected animal population is closer to the actual situation, which is used for accurate evaluation of the ability of natural enemies of pests in the subsequent period.

[0075] Step S4) predicting the type of pest 31 and the size of the pest according to the water body position 21, the terrain type 22, the crop type 24, the temperature data and the precipitation data.

[0076] The method for predicting the type of pest 31 and the size of the pest according to the water body position 21, the terrain type 22, the crop type 24, the temperature data and the precipitation data includes:

[0077] According to the water body position 21, the terrain type 22 and the precipitation data, the estimated value of soil humidity is calculated and obtained;

[0078] According to the soil humidity estimation value and temperature data, combined with the read pest suitable breeding environment information, all pest types 31 whose estimated population size exceeds the preset reference value are obtained;

[0079] According to the received pest-crop relationship information, which records the harm relationship of pests to crops, pest types 31 that are harmless to crops are screened out according to the crop type 24;

[0080] According to the remaining pest types 31, the predicted pest types 31 and pest size are obtained.

[0081] Based on the obtained water body position 21, terrain type 22 and precipitation data, the water accumulation and drainage characteristics in the region are analyzed, and the soil humidity level of the target region is estimated. For example, areas close to water bodies, low-lying or clay soils are more likely to maintain high humidity, while sloping or sandy soil areas drain faster and have lower humidity. Artificial measurement of the corresponding soil humidity can be used to mark the relevant data as sample data, and a machine learning model can be trained using sample data. Combined with the soil humidity estimation value and temperature data, the pre-stored pest suitable breeding environment information is compared. This information records the adaptability of different pests to environmental factors such as temperature and humidity, light, and soil conditions. Pests that are suitable for breeding under current environmental conditions and whose estimated population size may exceed the set reference value are selected to form a candidate list. The pest-crop relationship information describes the host relationship and harm relationship between various pests and different crops. According to the actual crop type 24 in the current region, those pests that are not the host of the crop and have no harm ability are removed. The remaining pest types 31 are the predicted pest types 31 that may cause harm in the region, and their population size is quantitatively evaluated in combination with environmental suitability.

[0082] For example, in a rice planting area, it is obtained that there are rivers and irrigation channels nearby, the terrain type 22 is low-lying plain, the crop type 24 is rice, and the recent temperature is 26-30°C per day and the precipitation data is continuous rainfall with a cumulative precipitation of 80mm. It is judged that the area has poor drainage and the soil is in a long-term high humidity state, so the soil humidity estimation value is calculated as "high humidity". The "high temperature + high humidity" environmental conditions are matched with the "pest suitable breeding environment information". In combination with the fact that rice planthoppers reproduce rapidly in a temperature of 25-30°C and high humidity environment, and the estimated population size will significantly exceed the reference value, corn borers, although they prefer warm environments and have a wide range of soil humidity adaptation, also have certain breeding potential in this environment, so they are also included in the candidate pest list.

[0083] Then, the pest-crop relationship information is read to obtain that rice planthopper is the main piercing-sucking pest of rice and can cause serious damage, while the corn borer can occasionally infest rice but its main damage object is corn and the damage in pure rice field is minimal. Since the current regional crop is rice and no corn is planted, the corn borer is excluded and only the rice planthopper is retained. The rice planthopper is determined as the predicted pest type 31 of the region, and its pest scale is quantified as "large-scale occurrence risk" according to the environmental suitability.

[0084] Step S5) The pest scale is corrected according to the animal population, and the prediction result of crop pest is obtained according to the corrected pest scale and the crop type 24.

[0085] The food chain model is read, which records the balanced proportion of the animal population and the pest scale;

[0086] The balanced number of the animal population of each species is calculated according to the pest scale;

[0087] The ratio of the balanced number to the estimated number of the animal population of each species is calculated, and the average value of the ratios corresponding to all the animal populations is taken as a correction coefficient;

[0088] The pest scale is corrected using the correction coefficient to obtain the corrected pest scale;

[0089] The tolerance limit of the pest scale for each crop type 24 is established, and when the pest scale exceeds the tolerance limit, the pest warning of the crop is generated.

[0090] A preset food chain model is read, which records the predatory relationship between different animal populations (such as birds, spiders, frogs, and other natural enemies) and a specific pest and the scale balance ratio thereof in an ecological balance state, i.e., the number of pests that can be effectively controlled by a unit number of natural enemies. In specific implementation, a region with similar climate, i.e., similar ecology, can be obtained, observation can be performed, and a statistical mean of the number of animal populations and pests can be obtained by using a statistical technique to obtain the scale balance ratio, and then a food chain model can be constructed. According to a currently predicted pest scale, in combination with the ratio relationship in the food chain model, a balance number of each type of natural enemy animal population that should be reached in an ideal balance state of the ecological system is reversely calculated. The balance number is compared with the corrected animal population prediction number to calculate a "balance number / prediction number" ratio of each type of natural enemy, which reflects whether the current number of natural enemies is sufficient to control the pests. An average value of the ratios of all related animal populations is obtained as a comprehensive correction coefficient. If the number of natural enemies is insufficient, the correction coefficient is greater than 1; if the number of natural enemies is sufficient, the coefficient is less than 1. The original predicted pest scale is weighted and adjusted by using the correction coefficient to obtain a corrected pest scale, so that the corrected pest scale is more consistent with the actual ecological control capability. According to a set pest scale tolerance limit value (i.e., the maximum pest density that can be tolerated by a crop without causing significant yield reduction) of each type of crop, it is determined whether the corrected pest scale exceeds the threshold value. If the corrected pest scale exceeds the threshold value, a corresponding crop pest warning is generated to indicate that there is a high risk and that prevention and control measures need to be taken. Thus, a closed-loop prediction from ecological correction to pest warning is completed.

[0091] For example, in a certain rice planting area, the initial prediction of the rice planthopper pest scale is 500 per mu of rice, which has a high risk of occurrence. In order to improve the prediction accuracy, the natural enemy animal population is introduced for ecological correction. A built-in food chain model is read to obtain that about 100 rice planthoppers can be effectively controlled by 10 rice field spiders or 2 sparrows in an ecological balance state. According to the current pest scale, it is calculated that, in theory, about 500 rice field spiders and about 50 sparrows are needed to maintain ecological balance, which is the balance number of each type of natural enemy. Through the identification of the previous aerial image 12 and the population prediction correction, it is obtained that the actual estimated number of natural enemies in this area is 300 rice field spiders and 30 sparrows. The ratios are calculated as follows: 500 / 300≈1.67 for spiders and 50 / 30≈1.67 for sparrows. The average value of the ratios is obtained as a correction coefficient of 1.67, which indicates that the number of natural enemies is insufficient, and the pest scale should be increased. The original prediction value is corrected by using the coefficient to obtain a corrected pest scale of 500x1.67≈835 per mu. The pest tolerance limit value of rice is obtained, which is set to be 600 rice planthoppers per mu. Since the corrected value is 835, which exceeds the limit value, it is determined that the risk is upgraded, and a red warning of rice planthopper pest is generated to indicate that prevention and control should be immediately carried out. By introducing the food chain relationship and the actual number of natural enemies, the pest prediction is closer to the real ecological condition, and the scientificity and reliability of the warning are improved.

[0092] In another aspect, the present specification provides a dispersed planting crop pest prediction system based on multi-modal data, please refer to the accompanying drawings Figure 4 , comprising:

[0093] The reading module 100 reads the map 11 of the target area, obtains the water body position 21 and the terrain type 22 according to the map 11, and reads the temperature data and precipitation data within a preset time period of the target area;

[0094] The identification module 200 reads the aerial image 12 of the target area within a preset time period, and identifies the crop type 24, the plant population 25 and the animal individual 23 in the target area;

[0095] The correction module 300 corrects the predicted animal population according to the plant population 25, and uses the animal individual 23 to correct the predicted animal population;

[0096] The prediction module 400 predicts the pest type 31 and the pest scale according to the water body position 21, the terrain type 22, the crop type 24, the temperature data and the precipitation data;

[0097] The result module 500 corrects the pest scale according to the animal population, and obtains the prediction result of crop pest according to the corrected pest scale and the crop type 24.

[0098] Please refer to Figure 5 The structure schematic diagram of the electronic device provided by the embodiment of the present specification is shown.

[0099] As Figure 5As shown, the electronic device 1100 can include at least one processor 1101, at least one network interface 1104, a user interface 1103, a memory 1105, and at least one communication bus 1102. The communication bus 1102 can be used to realize the connection and communication of the above-mentioned components. The user interface 1103 can include a key, and the optional user interface can also include a standard wired interface, a wireless interface. The network interface 1104 can include, but is not limited to, a Bluetooth module, an NFC module, a Wi-Fi module, etc. The processor 1101 can include one or more processing cores. The processor 1101 connects various parts in the entire electronic device 1100 through various interfaces and lines, executes various functions of the routing device and processes data by running or executing instructions, programs, code sets or instruction sets stored in the memory 1105, and calling data stored in the memory 1105. Optionally, the processor 1101 can be realized by at least one of DSP, FPGA, PLA. The processor 1101 can integrate CPU, GPU, modem and the like. Among them, the CPU mainly processes operating systems, user interfaces and application programs; the GPU is responsible for rendering and drawing the content to be displayed on the display screen; and the modem is used for processing wireless communication.

[0100] It can be understood that the above-mentioned modem can also not be integrated into the processor 1101, but be realized by a separate chip.

[0101] The memory 1105 can include RAM and ROM. Optionally, the memory 1105 includes a non-transitory computer readable medium. The memory 1105 can be used to store instructions, programs, codes, code sets or instruction sets. The memory 1105 can include a program storage area and a data storage area. The program storage area can store instructions for implementing an operating system, instructions for at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing the above-mentioned various method embodiments, etc. The data storage area can store data involved in the above-mentioned various method embodiments, etc. The memory 1105 can also be at least one storage device located away from the above-mentioned processor 1101. The memory 1105 as a computer storage medium can include an operating system, a network communication module, a user interface module and an application program. The processor 1101 can be used to call the application program stored in the memory 1105 and execute the method in the above-mentioned multiple embodiments.

[0102] The embodiments of the present specification also provide a computer-readable storage medium, which stores instructions, when the instructions are executed on a computer or a processor, cause the computer or the processor to perform the steps of the above-mentioned embodiments. The constituent modules of the above-mentioned electronic device, if realized in the form of software function units and sold or used as independent products, can be stored in the computer-readable storage medium.

[0103] The embodiments of the present specification also provide a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of the above-mentioned embodiments.

[0104] In the case of no conflict, the technical features in the embodiments and the implementation forms can be combined arbitrarily.

[0105] In the above-mentioned embodiments, all or part of the embodiments can be realized by software, hardware, firmware or any combination thereof. When realized by software, all or part of the embodiments can be realized in the form of a computer program product. The computer program product comprises a plurality of computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of the present specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable device. The computer instructions can be stored in or transmitted by a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired (such as coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (such as infrared, wireless, microwave, etc.) manner. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center, etc. integrated with a plurality of available media. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a digital versatile disc (DVD)), or a semiconductor medium (such as a solid state disk (SSD)) and the like.

[0106] When implemented by hardware or firmware, the foregoing method flow is programmed into a hardware circuit to obtain a corresponding hardware circuit structure, and the corresponding function is realized. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is an integrated circuit, and the logic function thereof is determined by the user programming the device. A digital system is "integrated" on a PLD by the designer programming it by himself / herself, without the need to ask a chip manufacturer to design and manufacture a special integrated circuit chip. Moreover, nowadays, instead of manually manufacturing an integrated circuit chip, this programming is mostly implemented by using a "logic compiler" software, which is similar to a software compiler used when a program is developed and written, and the original code before the compilation also needs to be written in a specific programming language, which is called a hardware description language (HDL), and there are many kinds of HDLs. Those skilled in the art should understand that, as long as the method flow is logically programmed in the above-mentioned several hardware description languages and programmed into an integrated circuit, a hardware circuit that realizes the logical method flow can be easily obtained.

[0107] The above-described embodiments are merely preferred embodiments of the present specification, and do not limit the scope of the present specification. Various changes and modifications to the technical solutions of the present specification made by those skilled in the art without departing from the design spirit of the present specification shall fall within the protection scope of the claims of the present specification.

Claims

1. A method for predicting pests in dispersedly planted crops based on multi-modal data, characterized in that, Including the following steps: Read the map of the target area, obtain the location of water bodies and terrain types based on the map, and read the temperature data and precipitation data of the target area within a preset time period; Read aerial images of the target area within a preset time period and identify crop types, plant populations, and individual animals within the target area; The animal population is predicted based on the plant population, and the predicted animal population is corrected using the individual animals. Based on the location of water bodies, topography, crop type, temperature data, and precipitation data, predict the types and scale of pests. The pest size is adjusted based on the animal population, and crop pest prediction results are obtained based on the adjusted pest size and crop type.

2. The method for predicting pests in dispersed crops based on multi-model data according to claim 1, characterized in that, The map is a GIS map. Methods for reading a map of a target area and obtaining the location of water bodies and terrain types based on the map include: Read the land use and cover map, and obtain the location and extent of water bodies and vegetation based on the land use and cover map; Read the digital elevation model and obtain the location and extent of the mountain based on the digital elevation model; Based on the location and extent of the mountains, water bodies, and vegetation, the terrain type of the target area is obtained.

3. The method for predicting pests in dispersed crops based on multi-model data according to claim 1 or 2, characterized in that, Methods for reading aerial images of a target area within a preset time period and identifying crop types, plant populations, and individual animals within the target area include: Calculate the off-axis angle of each pixel in the aerial image, delete aerial image regions with off-axis angles greater than a preset threshold, perform image recognition on the remaining aerial image regions, and divide the aerial image into image blocks according to the recognized objects; Obtain the shooting location and camera parameter information of the aerial image, align the image patch with the map of the target area, and obtain the shooting time, location marker, and shooting angle marker of the image patch; Based on the objects, location markers, and shooting angle markers identified from all image patches, the crop type and region within the target area are obtained; Based on the objects and location markers identified in all image patches, duplicate plant populations are aggregated to obtain the plant populations within the target area. Based on the objects identified from all image blocks, the shooting time, location markers, and shooting angle markers, individual animals within the target area are obtained, and the location and time of appearance of the individual animals are marked.

4. The method for predicting pests in dispersed crops based on multi-model data according to claim 3, characterized in that, The method for predicting animal populations based on plant populations and refining the predicted animal populations using individual animals includes: Receive plant and animal ecological association information, wherein the plant and animal ecological association information records the habitat support capacity of plant populations for animal populations; Based on the distribution of plant populations within the target area, an initial animal population prediction result is generated by reasoning through the plant-animal ecological association information. The animal population prediction result includes the species, spatial distribution probability, and estimated number of animal populations. Based on the animal population prediction results, animal habits, and plant populations, the estimated number distribution of individual animals photographed at each time period is obtained; The identified animal individuals, their locations, and times of occurrence are compared with the estimated population distribution to obtain a correction ratio coefficient. The predicted animal population is corrected using the aforementioned correction scaling factor.

5. The method for predicting pests in dispersed crops based on multi-model data according to claim 1 or 2, characterized in that, Methods for predicting pest types and scale based on water body location, terrain type, crop type, temperature data, and precipitation data include: Based on the location of the water body, the terrain type, and the precipitation data, the estimated soil moisture value is calculated. Based on the estimated soil moisture and temperature data, combined with the information on suitable breeding environments for pests, all pest types whose estimated population size exceeds the preset reference value are obtained. Based on the received pest-crop relationship information, which records the damage relationship between pests and crops, pest types that do not cause harm are screened out according to crop type. Based on the remaining pest types, predict the pest types and pest scale.

6. The method for predicting pests in dispersed crops based on multi-model data according to claim 5, characterized in that, The method for correcting the pest size based on the animal population, and obtaining crop pest prediction results based on the corrected pest size and crop type includes: Read the food chain model, which records the size balance ratio between animal populations and pests; Calculate the balance population size of each animal species based on the scale of the pests. Calculate the ratio of the equilibrium population size to the estimated population size for each animal species, and use the average of the ratios for all animal species as a correction factor. The pest size is corrected using the correction coefficient to obtain the corrected pest size. Establish a tolerance limit for pest size for each crop type. When the pest size exceeds the tolerance limit, generate a pest alarm for the crop.

7. A pest prediction system for decentralized crops based on multi-modal data, characterized in that, include: The reading module reads a map of the target area, obtains the location of water bodies and terrain types based on the map, and reads temperature and precipitation data for the target area within a preset time period. The identification module reads aerial images of the target area within a preset time period and identifies crop types, plant populations, and individual animals within the target area. The correction module predicts the animal population based on the plant population and corrects the predicted animal population using the individual animals. The prediction module predicts the type and scale of pests based on the location of water bodies, terrain type, crop type, temperature data, and precipitation data. The results module corrects the pest size based on the animal population and obtains crop pest prediction results based on the corrected pest size and crop type.

8. An electronic device, characterized in that, Including the processor and memory; The processor is connected to the memory; The memory is used to store executable program code; The processor runs a program corresponding to the executable program code stored in the memory to perform the method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-6.

Images