Gastrodia elata growth monitoring method and system based on image recognition

Abstract

The application provides a kind of ginseng growth monitoring method and system based on image recognition, related to ginseng growth monitoring technical field, the core step of the application includes dividing ginseng planting area into multiple grids, and collecting relevant environmental data through environmental sensor, using convolutional neural network to build growth monitoring model to obtain ginseng growth data, further analyze model output to obtain health index.The method can quickly identify the growth condition and potential disease of ginseng through accurate growth data analysis.The application not only provides a more comprehensive growth monitoring means, but also calculates the comprehensive growth index by comprehensively analyzing the health index, environmental data and planting density to realize the comprehensive and scientific evaluation of ginseng growth status.Through this method, the grower can timely make scientific decisions and optimize management measures to ultimately promote the yield and quality improvement of ginseng.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Gastrodia growth monitoring, in particular to a Gastrodia growth monitoring method and system based on image recognition. BACKGROUND

[0002] Gastrodia is a precious traditional Chinese medicine that has attracted widespread attention due to its unique medicinal and economic value. As people increasingly value health and wellness, the market demand for Gastrodia continues to grow, driving the expansion of its cultivation area. However, Gastrodia has a long growth cycle and strict environmental requirements, including changes in soil moisture, light intensity, and temperature, which directly affect its growth quality. Therefore, monitoring the growth status and health of Gastrodia is particularly important to ensure timely management measures and protect yield and quality.

[0003] Gastrodia growth is influenced by various factors such as environmental conditions, soil quality, and diseases, and changes in these factors directly affect the growth status and yield of Gastrodia. However, traditional monitoring methods mainly rely on manual observation and experience-based judgment, which is not only inefficient but also prone to subjective bias, making it difficult to provide scientific and accurate growth data. This results in the inability of growers to identify growth problems in a timely manner, affecting the quality and yield of Gastrodia. In addition, the lack of systematic data support also limits the effectiveness of scientific research and management decisions.

[0004] The existing technology has the following shortcomings:

[0005] Relying on manual inspection and traditional growth assessment methods, it is difficult for traditional methods to quickly identify problems in the face of complex growth environments and potential diseases, which affects the timely response of growers to the growth status of Gastrodia and increases the risk of disease spread and loss. In addition, existing technology often lacks systematic monitoring of environmental factors, making it difficult to analyze soil quality, climate change, and other variables closely related to Gastrodia growth. This one-sided monitoring method makes growers lack a comprehensive understanding of the growth environment of Gastrodia, further limiting the effectiveness of scientific assessment of growth conditions and management decisions. Therefore, the shortcomings of existing technology lie in the singularity and lag of its monitoring methods, which cannot meet the needs of modern agriculture for efficient, precise, and intelligent management, and have not effectively promoted the sustainable development of Gastrodia cultivation.

[0006] The above information disclosed in the background section is only used to enhance the understanding of the background of the present disclosure, and therefore it can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0007] The purpose of the present application is to provide a Gastrodia growth monitoring method and system based on image recognition to solve the problems raised in the background.

[0008] To achieve the above object, the present application provides the following technical solutions:

[0009] A ginseng growth monitoring method based on image recognition, the specific steps comprising:

[0010] Step 1: Divide the soil area where ginseng grows into several grids, collect the environmental data of each grid at the current time, and the environmental data includes soil humidity, temperature and light intensity;

[0011] Step 2: Obtain a plurality of ginseng growth images with known growth data, use ginseng growth images as input and corresponding growth data as output, and construct and train a growth monitoring model based on convolutional neural network;

[0012] Step 3: Divide all grids into a plurality of community sets according to the environmental data of each grid, select a current sampling area in each community set, and determine the planting density of the current sampling area based on the number of ginseng rhizomes in the current sampling area and the area of the sampling area;

[0013] Step 4: Obtain the growth image of ginseng in the current sampling area and input it into the growth monitoring model to obtain the growth data of ginseng in the current sampling area, and analyze the health status of ginseng in the current sampling area based on the growth data;

[0014] Step 5: According to the health status of ginseng in the current sampling area, the environmental data and the planting density in each community set, generate a comprehensive growth index to analyze the growth status of ginseng in each grid, and screen out the grids with abnormal growth status according to the comprehensive growth index.

[0015] Further, collecting the environmental data of each grid at the current time specifically comprises:

[0016] Divide the planting soil area of ginseng into a plurality of grids, and arrange soil humidity sensors, temperature sensors and light sensors at the center of the grid to collect the environmental data of the ginseng growth area.

[0017] Further, constructing and training the growth monitoring model specifically comprises:

[0018] Obtain a plurality of ginseng growth images with known growth data, each set of ginseng growth images includes images taken from various angles, and the growth data includes rhizome surface area, rhizome volume, and the proportion of the area where rhizome pests exist, the average rhizome surface area of each set of ginseng growth images is taken as the rhizome surface area of the corresponding ginseng, and the proportion is the ratio of the rhizome area where pests exist to the rhizome surface area, and the formula for calculating the proportion of the area where rhizome pests exist is:

[0019]

[0020] wherein, is the proportion of the area of the tuber on which the pests and diseases exist, is the surface area of the area of the tuber on which the pests and diseases exist, is the surface area of the tuber;

[0021] The growth data of each group of ginseng growth images is labeled, and each group of labeled ginseng growth images is preprocessed, including removing blurred and repeated images, and randomly rotating, flipping, cropping and adjusting the brightness of the images, and finally unifying the image size;

[0022] The labeled and preprocessed ginseng growth images are taken as input, a convolutional neural network is selected as the basic model, the corresponding growth data is taken as the label, the convolutional neural network model is trained, and a growth monitoring model is constructed.

[0023] Further, determining the planting density of the grid specifically includes:

[0024] Each grid is taken as a node, and the environmental data corresponding to each node is taken as the attribute of the node, a undirected graph is constructed, an edge is added in the undirected graph if two nodes are adjacent, and a undirected graph network is formed;

[0025] Each node is placed in an independent community, for each node in the undirected graph network, the node is moved to the community in which all neighbor nodes of the node are located, and the environmental change increment corresponding to each movement is calculated, specifically including:

[0026] First, the mean value of the environmental data between a node and any neighbor node of the node is calculated, and then the change between the node and the mean value is analyzed to obtain the environmental change increment, and the calculation formula is:

[0027]

[0028] wherein, is the environmental change increment of the i-th node and the j-th node, is the mean value of the soil moisture of the i-th node and the j-th node, is the soil moisture of the i-th node, is the mean value of the temperature of the i-th node and the j-th node, is the temperature of the i-th node, is the mean value of the light intensity of the i-th node and the j-th node, is the light intensity of the i-th node, i and j are nodes in the undirected graph, and i and j are neighbor nodes;

[0029] The node is moved to the neighbor community with the minimum environment change value, and the node is repeatedly moved until the community attribution of all nodes no longer changes, at this time, a node in each community is randomly selected as the current sampling area corresponding to all nodes in the community set, the planting density of the current sampling area is obtained, and the planting density formula is:

[0030]

[0031] wherein, is the planting density of the current sampling area, is the number of tubers in the current sampling area, is the area of the current sampling area;

[0032] The planting density of the current sampling area is regarded as the planting density of the community set in which the current sampling area is located, and the growth area of the Gastrodia elata Blume can be classified by the planting density: if , it is determined as a no-Gastrodia elata Blume area, if , it is determined as a sparse planting area, if , it is determined as a lush planting area, and if , it is determined as an over-dense planting area; when the growth area of the Gastrodia elata Blume is in the no-Gastrodia elata Blume area and the sparse planting area, a signal of increasing planting of the Gastrodia elata Blume in the grid is sent to the planting personnel, and when the growth area of the Gastrodia elata Blume is in the over-dense planting area, a signal of reducing planting of the Gastrodia elata Blume in the grid is sent to the planting personnel, is a sparse threshold value, is a dense threshold value.

[0033] Further, the analysis of the health condition of the Gastrodia elata Blume in the current sampling area specifically includes:

[0034] The growth image of the Gastrodia elata Blume in the current sampling area is obtained, input into the trained growth monitoring model, the growth data of the Gastrodia elata Blume in the current sampling area output by the model is obtained, the health index is generated according to the growth data of the Gastrodia elata Blume, so as to evaluate the health condition of the Gastrodia elata Blume in the current sampling area, and the calculation formula is:

[0035]

[0036] wherein, is the health index, is the tuber size, V is the tuber volume, is the proportion of the area with pests and diseases on the tuber, are weight coefficients of each item, , and ; the health index of the Gastrodia elata Blume in the current sampling area is regarded as the health index of the grid in which the current sampling area is located.

[0037] Further, the analysis of the environment data specifically includes:

[0038] According to the environmental data of each current sampling region, an environmental comprehensive index is calculated to evaluate the environment in the ginseng growing soil region, and the calculation formula is:

[0039]

[0040] wherein, is the environmental comprehensive index, is the soil humidity of the current sampling region, is the temperature of the current sampling region, is the light intensity of the current sampling region, is the maximum suitable value of the soil humidity, temperature and light intensity, is the weight coefficient of the soil humidity, temperature and light intensity, .

[0041] Further, the current sampling region with abnormal growth condition is screened out, and the screening method comprises:

[0042] The formula for calculating the comprehensive growth index of each current sampling region is:

[0043]

[0044] wherein, is the comprehensive growth index, is the health index, is the planting density of the current sampling region, is the weight coefficient of the health index, environmental comprehensive index and planting density of ginseng, , and ;

[0045] The comprehensive growth index of each current sampling region is compared with a preset growth threshold value. If the comprehensive growth index of the current sampling region at the current time is higher than the preset growth threshold value, it indicates that the ginseng growth condition of the current sampling region is good, i.e. all the grids in the community set corresponding to the current sampling region are in a good ginseng growth state, and then normal monitoring is maintained. If the comprehensive growth index of the current sampling region at the current time is lower than the preset growth threshold value, it indicates that the ginseng growth condition of the current sampling region is poor, i.e. all the grids in the community set corresponding to the current sampling region are in a poor ginseng growth state, and then all the grids in the community set of the current sampling region are marked as abnormal growth grids, and a warning of poor growth condition is issued. For the grids marked as abnormal growth, the environmental comprehensive index of the adjacent grids is also determined. If the environmental comprehensive index of an adjacent grid is lower than a preset environmental threshold value, the adjacent grid is also marked as abnormal, and a warning of poor growth condition is issued.

[0046] The application further provides a ginseng growth monitoring system based on image recognition, which is used to realize the ginseng growth monitoring method based on image recognition.

[0047] The data acquisition module is configured to divide the soil area where the ginseng grows into a plurality of grids, and acquire environmental data of each grid at the current time, the environmental data including soil humidity, temperature and light intensity.

[0048] The model construction module is configured to acquire ginseng growth images of a plurality of known growth data, take the ginseng growth images as input and the corresponding growth data as output, and construct and train a growth monitoring model based on a convolutional neural network.

[0049] The planting density analysis module is configured to divide all the grids into a plurality of community sets according to the environmental data of each grid, select a current sampling area in each community set, and determine the planting density of the current sampling area based on the number of ginseng tubers in the current sampling area and the area of the sampling area.

[0050] The real-time monitoring module is configured to acquire a growth image of the ginseng in the current sampling area, input the growth image into the growth monitoring model to obtain growth data of the ginseng in the current sampling area, and analyze the health condition of the ginseng in the current sampling area based on the growth data.

[0051] The comprehensive analysis module is configured to generate a comprehensive growth index based on the health condition of the ginseng in the current sampling area, the environmental data and the planting density in each community set, analyze the growth condition of the ginseng in each grid based on the comprehensive growth index, and screen out grids with abnormal growth conditions according to the comprehensive growth index.

[0052] In the above technical solution, the application provides the following technical effects and advantages:

[0053] The ginseng growth monitoring method based on image recognition can monitor the growth of ginseng in real time and accurately, identify diseases and growth abnormalities in time, and has the beneficial effects of significantly improving the efficiency and accuracy of monitoring, enabling the grower to make quick responses based on scientific data, thereby optimizing management measures and ensuring the healthy growth of ginseng. In addition, by comprehensively analyzing the growth data and environmental parameters of ginseng, the method provides comprehensive decision support for the grower, promotes the scientific and precise planting of ginseng, and provides a guarantee for improving the overall yield and quality. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 The figure is a schematic diagram of the overall method of the application;

[0055] Figure 2The system structure diagram of the present application. DETAILED DESCRIPTION

[0056] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to specific embodiments.

[0057] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present application should be understood as the general meaning understood by those skilled in the art to which the present application belongs. The terms "first", "second" and similar terms used in the present application do not represent any order, quantity or importance, but are only used to distinguish different components. The terms "include" or "contain" and similar terms mean that the elements or objects before the terms cover the elements or objects listed after the terms and their equivalents, and do not exclude other elements or objects. The terms "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. The terms "up", "down", "left", "right" and the like are only used to represent relative positional relationships, and when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0058] Embodiment:

[0059] Please refer to Figure 1 The present application provides a technical solution:

[0060] A ginseng growth monitoring method based on image recognition, the specific steps comprising:

[0061] Step 1: Divide the soil area where ginseng grows into several grids, collect the environmental data of each grid at the current time, and the environmental data includes soil humidity, temperature and light intensity;

[0062] In this embodiment, collecting the environmental data of each grid at the current time specifically includes:

[0063] The planting soil area of ginseng is evenly divided into multiple grids, and soil humidity sensors, temperature sensors and light sensors are arranged at the center of the grids to collect the environmental data of the ginseng growth area, and the environmental data includes soil humidity, temperature (here, temperature refers to environmental temperature) and light intensity.

[0064] The grid division method effectively realizes the meticulous management of the ginseng planting area, so that each grid can independently collect data, providing more accurate and localized information for subsequent analysis. This method can help the grower to identify growth abnormalities and environmental changes in a specific area in a timely manner and make targeted adjustments quickly.

[0065] Step 2: Obtain a plurality of Gastrodia elata growth images with known growth data, use the Gastrodia elata growth images as input and the corresponding growth data as output, and construct and train a growth monitoring model based on a convolutional neural network;

[0066] In this embodiment, the growth monitoring model is constructed and trained, specifically including:

[0067] A plurality of Gastrodia elata growth images with known growth data are obtained, each set of Gastrodia elata growth images including images taken from various angles, and each angle can be an angle taken from four front directions and two up and down directions. The growth data includes rhizome surface area, rhizome volume, and the proportion of the area of the rhizome where pests and diseases exist. The average rhizome surface area of each set of Gastrodia elata growth images is used as the rhizome surface area of the corresponding Gastrodia elata. The proportion refers to the ratio of the rhizome surface area where pests and diseases exist to the rhizome surface area. The formula for calculating the proportion of the area of the rhizome where pests and diseases exist is:

[0068]

[0069] wherein, is the proportion of the area of the rhizome where pests and diseases exist, is the surface area of the area of the rhizome where pests and diseases exist, is the rhizome surface area;

[0070] The growth data of each set of Gastrodia elata growth images is labeled, and the labeled each set of Gastrodia elata growth images is preprocessed, including removing blurred and duplicate images, and randomly rotating, flipping, cropping and adjusting the brightness of the images. Finally, the image size is unified.

[0071] The labeled and preprocessed Gastrodia elata growth images are used as input, a convolutional neural network is selected as the base model, and the corresponding growth data is used as the label. The convolutional neural network model is trained to construct the growth monitoring model.

[0072] The growth data of Gastrodia elata rhizome is measured manually, which provides reliable basic data for subsequent model training. This high-quality labeling can effectively improve the learning effect of the convolutional neural network, making the model more accurate and reliable in practical application. Secondly, the image preprocessing step ensures the consistency and effectiveness of the input data. By removing blurred and duplicate images, and randomly rotating, flipping, cropping and adjusting the brightness of the images, the model's ability to adapt to images under different conditions can be enhanced. This data augmentation technique not only improves the model's generalization ability and reduces the risk of overfitting, but also ensures that the model can accurately identify the growth characteristics of Gastrodia elata in real-world environments with varying lighting and angles.

[0073] Step 3: dividing all the grids into a plurality of community sets according to the environmental data of each grid, selecting a current sampling area in each community set, and determining the planting density of the current sampling area based on the number of rhizomes of Gastrodia elata in the current sampling area and the area of the sampling area;

[0074] In this embodiment, the determination of the planting density of the grid specifically comprises:

[0075] Each grid is taken as a node, and the environmental data corresponding to each node is taken as the attribute of the node, a non-directed graph is constructed, an edge is added in the non-directed graph if two nodes are adjacent, and a non-directed graph network is formed;

[0076] Each node is placed in an independent community, and for each node in the non-directed graph network, the node is moved to the community in which all neighbor nodes of the node are located, and the environmental change increment corresponding to each movement is calculated, which specifically comprises:

[0077] First, the mean value of the environmental data between a node and any neighbor node of the node is calculated, and then the change between the node and the mean value is analyzed to obtain the environmental change increment, and the calculation formula is:

[0078]

[0079] wherein, is the environmental change increment of the ith node and the jth node, is the mean value of the soil moisture of the ith node and the jth node, is the soil moisture of the ith node, is the mean value of the temperature of the ith node and the jth node, is the temperature of the ith node, is the mean value of the light intensity of the ith node and the jth node, is the light intensity of the ith node, and i and j are nodes in the non-directed graph, and i and j are neighbor nodes;

[0080] The node is moved to the neighbor community with the minimum environmental change increment, the node is repeatedly moved until the community attribution of all nodes no longer changes, at this time, a node in each community is randomly selected as the current sampling area corresponding to all nodes in the community set, and the planting density of the current sampling area is obtained, and the calculation formula of the planting density is:

[0081]

[0082] wherein, is the planting density of the current sampling area, is the number of rhizomes in the current sampling area, is the area of the current sampling area;

[0083] The planting density of the current sampling region is regarded as the planting density of the community set in which the current sampling region is located. The Gastrodia elata growth region can be classified by the planting density: if , it is determined as a Gastrodia elata-free region, if , it is determined as a sparse planting region, if , it is determined as a lush planting region, and if , it is determined as an over-dense planting region; when the Gastrodia elata growth region is in the Gastrodia elata-free region and the sparse planting region, a signal is sent to the planting personnel to increase planting of Gastrodia elata in the grid, and when the Gastrodia elata growth region is in the over-dense planting region, a signal is sent to the planting personnel to reduce planting of Gastrodia elata in the grid, is a sparse threshold value, is a dense threshold value.

[0084] The neighbor node refers to the grid regions corresponding to the two nodes being adjacent. The environmental change value specifically reflects the degree of change in environmental conditions between node i and node j. By calculating the difference between the mean values of soil moisture, temperature and light intensity and the node's own attributes, the potential impact of environmental change on plant growth can be quantified. The greater the value of the environmental change value, the more obvious the environmental difference between node i and its neighbor node j, which may mean that node i is not suitable as a sampling region under the current environmental conditions, prompting the need to consider moving to a more suitable neighbor community. Here, the purpose is to determine the current sampling region, and therefore, the smaller the environmental change value between adjacent grids should be, indicating that the environmental data difference between the two grids is small. Analyzing the region with a small environmental difference as a sampling region can more accurately represent the conditions of other grids to minimize the difference between different grids. The smaller the difference in soil moisture, temperature and light intensity between different nodes (grids), the smaller the environmental change value, and vice versa. The greater the difference in soil moisture, temperature and light intensity between different nodes (grids), the greater the environmental change value, and the environmental data difference between the grids is positively correlated with the environmental change value.

[0085] According to different planting density classifications (Gastrodia elata-free region, sparse planting region, lush planting region and over-dense planting region), the planting personnel can timely adjust the planting strategy. For example, a signal to increase planting is sent to the Gastrodia elata-free and sparse planting regions, which can improve the land utilization rate; and a signal to reduce planting is sent to the over-dense planting region, which can effectively prevent resource competition and plant diseases. The threshold values and The two thresholds should be set based on historical data, experimental results, and actual growth conditions in the region. They can be determined by statistically analyzing the planting density data over a period of time to find the corresponding distribution characteristics, or by referring to agricultural research literature to obtain research data on similar crop planting densities and comparing and analyzing them to determine the appropriate planting density standards for Gastrodia elata.

[0086] Step 4: Obtain the growth image of Gastrodia elata in the current sampling area and input it into the growth monitoring model to obtain the growth data of Gastrodia elata in the current sampling area, and analyze the health status of Gastrodia elata in the current sampling area based on the growth data;

[0087] In this embodiment, analyzing the health status of Gastrodia elata in the current sampling area specifically includes:

[0088] Obtain the growth image of Gastrodia elata in the current sampling area and input it into the trained growth monitoring model to obtain the growth data of Gastrodia elata in the current sampling area output by the model, generate a health index based on the growth data of Gastrodia elata, and evaluate the health status of Gastrodia elata in the current sampling area. The calculation formula is:

[0089]

[0090] wherein, is the health index, is the size of the tuber, V is the volume of the tuber, is the proportion of the area of the tuber where pests and diseases exist, are the weight coefficients of each item, , and .

[0091] The natural logarithm function is used to represent the influence of the size of the tuber, ensuring that it can handle smaller tuber areas. For example, when the tuber area is 0, a constant is added to avoid the undefined logarithm. The logarithmic function can effectively reflect the diminishing marginal utility of tuber size on health. As the size of the tuber increases, the contribution of unit area to health gradually decreases. For example, a small tuber with an area of 2 and a medium-sized tuber with an area of 10 , in this case, the increase from 2 to 10 may significantly improve the health index, because small tubers often lack nutrients and have poor growing conditions. However, if we increase from 10 to 20 , although the tuber is still increasing, the health benefits of the increase may not be as obvious as the increase from small to medium-sized tubers. The larger the size of the tuber, the greater the health index The higher, the more positive proportion. If the tuber area is small, the growth rate of the health index is relatively small. The same is true for the stem volume. Using a logarithmic function can capture the non-linear relationship between stem volume and health index, especially when the data shows a gradual increasing effect, which can more truly reflect the health improvement effect. In addition, logarithmic transformation can reduce the influence of large values on the overall health index, so that larger tubers or volumes do not excessively affect the calculation of the health index, keeping a more balanced assessment.

[0092] The higher the health index, the healthier the growth of Gastrodia elata, the better the overall growth of the tuber, and the less the influence of pests and diseases; on the contrary, a lower health index means that there may be growth problems or greater influence of pests and diseases. By establishing this health index, researchers and farmers can more intuitively assess the growth of Gastrodia elata, providing a basis for subsequent management and decision-making.

[0093] The size of the tuber is usually directly related to the growth and health of the plant, and larger tubers often mean that the plant has accumulated good nutrition, thus promoting healthy growth. Increased tuber size will lead to an increase in the health index. The volume of the tuber is related to the internal organization and nutrient reserves, and an increase in volume means that the growth potential and survival ability of the plant are improved, which is usually associated with a better health status. Accordingly, an increase in tuber volume will also increase the health index. The proportion of the disease and pest area is an important negative indicator of plant health. The larger the disease and pest area, the worse the health of the plant, affecting its growth and development, leading to a decrease in the health index. The use of a logarithmic function indicates the positive effect of tuber size on the health index. As the size of the tuber increases, the health index also increases, reflecting a positive role in promoting plant growth. Similarly, an increase in tuber volume increases the health index through a logarithmic function, reflecting the growth potential of the plant. The proportion of the disease and pest area is in the form of , showing its negative effect on the health index. As the proportion of disease and pests increases, the health index decreases, indicating that the health of the plant is threatened.

[0094] Diseases and pests are one of the key factors affecting the health of plants. The presence of diseases and pests directly leads to impaired plant growth, which can cause problems such as growth stagnation and reduced fruit quality. The severity of diseases and pests is often the primary indicator of plant health, so their weight should be relatively high. When the proportion of diseases and pests reaches a certain level, their impact on plant health can be dramatically increased, and even lead to plant death. Therefore, the change in the proportion of diseases and pests has a significant and crucial impact on the health index. Tubers size is still an important manifestation of plant growth, and larger tubers usually mean higher nutrient accumulation and growth potential. However, considering the direct threat of diseases and pests, the weight of tuber size can be appropriately reduced. Tuber volume is closely related to the nutrient reserves and growth potential of plants, and its importance lies in supporting healthy growth. However, compared to the negative impact of diseases and pests, the weight of tuber volume can still be slightly lower. Tuber volume is usually more significant than tuber size in terms of its impact on health. Tuber volume is the total volume of tubers, which is directly related to the nutrient storage capacity and growth potential of plants. Larger volume can improve the overall growth of plants, affecting their stress resistance, nutrient supply, and growth rate, so it is reasonable to give it a greater weight. This constraint ensures that the health index is a normalized indicator, ensuring that the total contribution of the weighted combination of multiple factors is 100%.

[0095] Step 5: Based on the current sampling area of each community set, the health status of Gastrodia elata, environmental data, and planting density, a comprehensive growth index is generated to analyze the growth status of each grid Gastrodia elata, and the grid with abnormal growth status is screened out according to the comprehensive growth index.

[0096] In this embodiment, analyzing the environmental data specifically includes:

[0097] According to the environmental data of each current sampling area, an environmental comprehensive index is calculated to evaluate the environment in the Gastrodia elata growth soil area. The calculation formula is:

[0098]

[0099] wherein, is the environmental comprehensive index, is the soil moisture of the current sampling area, is the temperature of the current sampling area, is the light intensity of the current sampling area, is the maximum suitable value of soil moisture, temperature, and light intensity, is the weight coefficient of soil moisture, temperature, and light intensity, ;

[0100] The maximum suitable value of soil humidity is set to 70% (volume of water), and too high humidity can cause root hypoxia, while too low humidity can affect the growth of Gastrodia elata. The suitable temperature for the growth of Gastrodia elata is generally between 15°C and 25°C. Too high or too low temperature can affect its growth and development. Too high temperature can cause increased water evaporation and even root burning, while too low temperature can cause slow growth, slow root growth, reduced ability to absorb water and nutrients, and the like. Therefore, the maximum suitable value of growth temperature is generally set to 20°C. Gastrodia elata is a shade-loving plant, and the suitable light intensity is generally between 2000 and 4000 Lux. Too strong light can cause the soil to dry, and too low light intensity can also affect the growth and development of Gastrodia elata, so the maximum suitable value of light intensity is set to 3000 Lux.

[0101] In this embodiment, the current sampling area with abnormal growth conditions is screened out, and the specific screening process includes:

[0102] The formula for calculating the comprehensive growth index of each current sampling area is:

[0103]

[0104] wherein, is the comprehensive growth index, is the health index, is the planting density of the current sampling area, is the weight coefficient of the health index, the environmental comprehensive index, and the planting density of Gastrodia elata, , and , is an adjustment coefficient;

[0105] The comprehensive growth index of each current sampling area is compared with the preset growth threshold value. If the comprehensive growth index of the current sampling area at the current time is higher than the preset growth threshold value, it indicates that the Gastrodia elata in the current sampling area has good growth conditions, i.e., all the grids in the community set corresponding to the current sampling area are in good growth state, and normal monitoring is maintained. If the comprehensive growth index of the current sampling area at the current time is lower than the preset growth threshold value, it indicates that the Gastrodia elata in the current sampling area has poor growth conditions, i.e., all the grids in the community set corresponding to the current sampling area are in poor growth state, and all the grids in the community set of the current sampling area are marked as abnormal growth grids, and a warning of poor growth conditions is issued. For the grids marked as abnormal growth, the environmental comprehensive index of the adjacent grids is also determined. If the environmental comprehensive index of an adjacent grid is lower than the preset environmental threshold value, the adjacent grid is also marked as abnormal, and a warning of poor growth conditions is issued.

[0106] The health index, presented in index form, effectively emphasizes its impact on the overall growth index. If the plant is in good health, the health index will be higher. If the value is high, then The value increases significantly, reflecting the crucial role of health status in the growth process. Using a logarithmic function for the comprehensive environmental index can smooth out the influence of environmental data and reduce the interference of extreme values ​​on the comprehensive growth index. A key feature of the logarithmic function is that its influence diminishes as environmental conditions improve; its advantage lies in ensuring that even with improved environmental data... Even when the value is 0, the exponent is not undefined. Furthermore, adding... It can be used as a weighting coefficient, which allows for flexible adjustment under different environmental conditions. This method effectively controls the impact of planting density, ensuring a higher proportion of growth index at low densities and saturation at high densities. This means that even with increased planting density, the growth rate of the overall growth index gradually decreases, reflecting the resource competition and pressure that excessive density may lead to. Adjustment coefficient The value range is [0,1]. Smaller values, such as 0.1 or 0.2, indicate that planting density has a smaller impact on the growth index, while larger values ​​(close to 1) indicate that planting density has a significant impact on growth. Here, a smaller value can be set to reflect that the impact of planting density on the overall growth index is relatively small.

[0107] Health Index It is usually the most important indicator reflecting the condition of a plant, and healthy plants play a central role in growth and physiological activities. Therefore, weighting It should be set to maximum to reflect its importance. Environmental conditions (such as soil moisture, temperature, and light) directly affect plant growth and health, but their impact on planting density may be more direct and obvious. Changes in environmental conditions can lead to significant changes in plant health, therefore... It can be set to greater than or equal to Planting density reflects the number of Gastrodia elata tubers in the area. Excessive planting density makes it difficult to meet the nutrient needs of the tubers in that area. Compared to health index and environmental data, its impact is relatively small, therefore its weight can be set relatively low. Therefore, here... This reflects that the higher the planting density, the lower the comprehensive impact index, and vice versa.

[0108] By comparing the comprehensive growth index with the preset growth threshold, the grid with abnormal growth status can be quickly identified. When the comprehensive growth index is lower than the threshold, the system can timely issue a warning to ensure that the grower can quickly take measures such as adjusting irrigation, fertilization or improving other environmental conditions. This timely response is crucial for the healthy growth of Gastrodia. After marking a grid as abnormal, the system also judges the environmental comprehensive index of the adjacent grid. This linkage monitoring mechanism can effectively prevent the spread of potential diseases or unfavorable growth conditions, ensuring the health management of the entire planting area. In addition, by paying attention to the adjacent grid, the management strategy can be adjusted in time to avoid large-scale losses. By establishing the relationship between the comprehensive growth index and the environment and plant physiology, the grower can make decisions based on scientific data rather than relying on traditional experience. This flexible management approach helps to continuously improve the production process and adapt to different growth conditions and market demands.

[0109] Referring to Figure 2 The application further provides a Gastrodia growth monitoring system based on image recognition, which is used to implement the above-mentioned Gastrodia growth monitoring method based on image recognition and comprises:

[0110] A data acquisition module is configured to divide the soil area where Gastrodia grows into a plurality of grids, acquire the environmental data of each grid at the current time, and the environmental data includes soil humidity, temperature and light intensity.

[0111] A model construction module is configured to acquire a plurality of Gastrodia growth images with known growth data, take the Gastrodia growth images as input and the corresponding growth data as output, construct and train a growth monitoring model based on a convolutional neural network.

[0112] A planting density analysis module is configured to divide all the grids into a plurality of community sets according to the environmental data of each grid, select a current sampling area in each community set, determine the planting density of the current sampling area based on the number of tubers of Gastrodia in the current sampling area and the area of the sampling area.

[0113] A real-time monitoring module is configured to acquire the growth image of Gastrodia in the current sampling area and input it into the growth monitoring model to obtain the growth data of Gastrodia in the current sampling area, and analyze the health status of Gastrodia in the current sampling area based on the growth data.

[0114] A comprehensive analysis module is configured to generate a comprehensive growth index based on the health status of Gastrodia in the current sampling area, the environmental data and the planting density in each community set to analyze the growth status of Gastrodia in each grid, and screen out the grid with abnormal growth status according to the comprehensive growth index.

[0115] The above formulas are all dimensionless values calculated, the formula is obtained by collecting a large amount of data to simulate the most recent real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0116] The above embodiments can be implemented wholly or partially by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented wholly or partially in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of the examples described in connection with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solutions.

[0117] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, which can be located in one place or distributed on multiple network units. Part or all of the units can be selected to achieve the purpose of the embodiments according to actual needs.

[0118] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered within the protection scope of the present application.

Claims

1. A ginseng growth monitoring method based on image recognition, characterized by, The specific steps include: Step 1: divide the soil area where the Gastrodia elata grows into a plurality of grids, collect the environment data of each grid at the current time, and the environment data includes soil humidity, temperature and light intensity; Step 2: obtain a plurality of Gastrodia elata growth images with known growth data, take the Gastrodia elata growth image as input and the corresponding growth data as output, and construct and train a growth monitoring model based on a convolutional neural network; Step 3: divide all the grids into a plurality of community sets according to the environment data of each grid, select a current sampling area in each community set, and determine the planting density of the current sampling area based on the number of rhizomes of Gastrodia elata in the current sampling area and the area of the sampling area; Step 4: obtain the growth image of Gastrodia elata in the current sampling area and input it into the growth monitoring model to obtain the growth data of Gastrodia elata in the current sampling area, and analyze the health status of Gastrodia elata in the current sampling area based on the growth data; Step 5: generate a comprehensive growth index according to the health status of Gastrodia elata in the current sampling area, the environment data and the planting density in each community set to analyze the growth status of Gastrodia elata in each grid, and screen out the grids with abnormal growth status according to the comprehensive growth index; The specific steps of determining the planting density of the grid include: Each grid is taken as a node, and the environment data corresponding to each node is taken as the attribute of the node, a undirected graph is constructed, an edge is added in the undirected graph if two nodes are adjacent, and a undirected graph network is formed; Each node is placed in an independent community, for each node in the undirected graph network, it is moved to the community where all its neighbor nodes are located, and the corresponding environmental change increment is calculated each time the node is moved, which specifically includes: Firstly, the mean value of the environment data between a node and any of its neighbor nodes is calculated, and then the change between the node and the mean value is analyzed to obtain the environmental change increment, and the calculation formula is: wherein, is the average of the environmental change value of the i-th node and the j-th node, is the average of the soil moisture of the i-th node and the j-th node, is the soil moisture of the i-th node, is the average of the temperature of the i-th node and the j-th node, is the temperature of the i-th node, is the average of the light intensity of the i-th node and the j-th node, is the light intensity of the i-th node, i, j are nodes in the undirected graph, and i and j are neighboring nodes; The node is moved to the neighbor community with the smallest environmental change increment, the node is repeatedly moved until the community attribution of all nodes no longer changes, at this time a node in each community is randomly selected as the current sampling area of all nodes in the community set, and the planting density of the current sampling area is obtained, and the calculation formula of the planting density is: wherein, is the planting density of the current sampling area, is the number of tubers in the current sampling area, is the area of the current sampling area; The planting density of the current sampling area is regarded as the planting density of the community set in which the current sampling area is located. The ginseng growing area can be classified by the planting density: if , it is determined as a no ginseng area; if , it is determined as a sparse planting area; if , it is determined as a flourishing planting area; if , it is determined as an over-dense planting area; when the ginseng growing area is in the no ginseng area and the sparse planting area, a signal is sent to the planting personnel that the grid should increase planting of ginseng; when the ginseng growing area is in the over-dense planting area, a signal is sent to the planting personnel that the grid should reduce planting of ginseng, is a sparse threshold value, is a dense threshold value.

2. The image recognition-based Gastrodia elata growth monitoring method according to claim 1, characterized in that, The specific steps of collecting the environment data of each grid at the current time include: The planting soil area of Gastrodia elata is evenly divided into a plurality of grids, soil humidity sensors, temperature sensors and light sensors are arranged at the centers of the grids to collect the environment data of the Gastrodia elata growth area.

3. The image recognition-based Gastrodia elata growth monitoring method according to claim 1, characterized in that, The specific steps of constructing and training the growth monitoring model include: Obtain a plurality of Gastrodia elata growth images with known growth data, each set of Gastrodia elata growth images includes images taken from various angles, the growth data includes rhizome surface area, rhizome volume, and the proportion of the area where the rhizome has pests and diseases, the mean value of the rhizome surface area of each set of Gastrodia elata growth images is taken as the rhizome surface area of the corresponding Gastrodia elata, and the proportion is the ratio of the rhizome area where pests and diseases exist to the rhizome surface area, and the calculation formula of the proportion of the area where the rhizome has pests and diseases is: wherein is the proportion of the area of the tuber on which the area of the pest and disease is present, is the surface area of the area of the pest and disease present in the tuber, is the surface area of the tuber; The growth data of each group of Gastrodia elata growth images is labeled, and the labeled growth images of each group of Gastrodia elata are preprocessed, including removing blurred and repeated images, and randomly rotating, flipping, cropping and adjusting the brightness of the images, and finally unifying the image size; The labeled and preprocessed Gastrodia elata growth images are taken as inputs, a convolutional neural network is selected as a basic model, the corresponding growth data is taken as labels, the convolutional neural network model is trained, and a growth monitoring model is constructed.

4. The image recognition-based Gastrodia elata growth monitoring method according to claim 1, characterized in that, The analysis of the health status of Gastrodia elata in the current sampling area specifically includes: The growth image of Gastrodia elata in the current sampling area is obtained, input into the trained growth monitoring model, the growth data of Gastrodia elata in the current sampling area output by the model is obtained, and a health index is generated according to the growth data of Gastrodia elata to evaluate the health status of Gastrodia elata in the current sampling area, and the calculation formula is: wherein, is a health index, is a tuber size, V is a tuber volume, is a proportion of an area where a disease or pest exists on a tuber, are weight coefficients of each item, respectively, , and ; a health index of the Gastrodia elata in the current sampling area is regarded as a health index of the grid where the current sampling area is located.

5. The image recognition based Gastrodia elata growth monitoring method according to claim 1, characterized in that, The analysis of the environment data specifically includes: According to the environment data of each current sampling area, an environment comprehensive index is calculated to evaluate the environment in the Gastrodia elata growth soil area, and the calculation formula is: wherein, is an environmental comprehensive index, is the soil humidity of the current sampling area, is the temperature of the current sampling area, is the light intensity of the current sampling area, is the maximum suitable value of the soil humidity, temperature, and light intensity, is the weight coefficient of the soil humidity, temperature, and light intensity, .

6. The image recognition-based Gastrodia elata growth monitoring method according to claim 5, characterized in that, The current sampling area with abnormal growth condition is screened out, and the calculation formula of the comprehensive growth index of each current sampling area is: The comprehensive growth index of each current sampling area is compared with a preset growth threshold value, if the comprehensive growth index of the current sampling area at the current time is higher than the preset growth threshold value, it indicates that the growth condition of Gastrodia elata in the current sampling area is good, that is, all the grids in the community set corresponding to the current sampling area are in a good growth state of Gastrodia elata, and then normal monitoring is maintained, if the comprehensive growth index of the current sampling area at the current time is lower than the preset growth threshold value, it indicates that the growth condition of Gastrodia elata in the current sampling area is poor, that is, all the grids in the community set corresponding to the current sampling area are in a poor growth state of Gastrodia elata, then all the grids in the community set of the current sampling area are marked as abnormal growth grids, and a warning of poor growth condition is issued, and for the grids marked as abnormal growth, the environment comprehensive index of the adjacent grids is also judged, if the environment comprehensive index of an adjacent grid is lower than a preset environment threshold value, the adjacent grid is also marked as abnormal, and a warning of poor growth condition is issued. wherein, is a comprehensive growth index, is a health index, is a planting density of a current sampling area, is a weight coefficient of the health index, the comprehensive environmental index, and the planting density of the Gastrodia elata, , and , is an adjustment coefficient; The Gastrodia elata growth monitoring system based on image recognition is used to implement the Gastrodia elata growth monitoring method based on image recognition in any one of claims 1-6, and includes:

7. An image recognition-based Gastrodia elata growth monitoring system, characterized by, A data acquisition module is configured to divide a soil area where Gastrodia elata grows into a plurality of grids, and acquire environment data of each grid at a current time, the environment data including soil humidity, temperature and light intensity; A model construction module is configured to acquire a plurality of Gastrodia elata growth images with known growth data, take the Gastrodia elata growth images as inputs, and take the corresponding growth data as outputs, and construct and train a growth monitoring model based on a convolutional neural network; A planting density analysis module is configured to divide all the grids into a plurality of community sets according to the environment data of each grid, select a current sampling area in each community set, determine the planting density of the current sampling area based on the number of tubers of Gastrodia elata in the current sampling area and the area of the sampling area. ​ A real-time monitoring module is configured to acquire a growth image of the Gastrodia elata in the current sampling area and input the growth image into a growth monitoring model to obtain growth data of the Gastrodia elata in the current sampling area, and analyze the health condition of the Gastrodia elata in the current sampling area based on the growth data; A comprehensive analysis module is configured to generate a comprehensive growth index based on the health condition of the Gastrodia elata in the current sampling area, the environmental data and the planting density in each community set to analyze the growth condition of the Gastrodia elata in each grid, and screen out the grids with abnormal growth conditions according to the comprehensive growth index.

Images