A method and system for rapid detection of high oil corn seedling plant number

Abstract

The application belongs to the field of agricultural information technology, and discloses a kind of for high oil corn seedling stage plant number Rapid detection method and system, the method includes the following steps: selecting high oil corn monitoring field block;Through unmanned aerial vehicle carrying image acquisition equipment, obtain high oil corn field multidimensional data;Multidimensional data preprocessing;Seedling segmentation and identification;Plant number statistics and verification;Data storage and analysis.The application adopts the above-mentioned rapid detection method and system for high oil corn seedling stage plant number, which provides data support for the analysis and prediction of corn growth conditions, helps to optimize field management measures, and improves corn yield and quality;It can quickly and accurately detect the plant number of high oil corn seedling stage, improve the efficiency and accuracy of field management;It has wide applicability and can be used for corn field management in different growth stages and growth environments.

Description

Technical Field

[0001] This invention relates to the field of agricultural information technology, and in particular to a rapid detection method and system for the number of seedlings in high-oil corn. Background Technology

[0002] Traditional methods for detecting the number of corn seedlings rely on manual counting, which is inefficient and prone to errors. While current UAV remote sensing and computer vision technologies have made progress in crop monitoring—for example, threshold segmentation based on RGB imagery achieves an accuracy of 89%-96%, and the YOLOv8 model achieves 98.22% accuracy in identifying ordinary corn seedlings—significant limitations remain for high-oil corn.

[0003] 1. Poor environmental adaptability: The glossy reflective properties of high-oil corn leaves cause the accuracy of traditional image segmentation methods to drop sharply by 6%-10% on cloudy days. The low contrast between seedlings and weeds and stubble further increases the false detection rate.

[0004] 2. Insufficient model generalization: Most existing deep learning models are trained on common corn and lack optimization for the morphological differences of high-oil corn (narrow leaf sheaths, dark green leaves), and parameters need to be adjusted according to the growth stage.

[0005] 3. Hardware deployment bottleneck: High-precision models (such as Faster R-CNN) rely on GPU computing power, while lightweight models (such as YOLO-FE2) have a false negative rate of >15% under high-density planting.

[0006] 4. Lack of agronomic integration: The existing system does not link key agronomic parameters such as seedling uniformity and oil content, which restricts the value of decision-making.

[0007] Therefore, there is an urgent need to develop a rapid detection system for the number of high-oil maize seedlings that combines environmental robustness, dedicated model optimization, and embedded deployment capabilities to fill the technological gap. Summary of the Invention

[0008] The purpose of this invention is to provide a rapid detection method and system for the number of seedlings in high-oil maize. This method and system combine modern sensor technology and image processing technology to achieve rapid and accurate detection of the number of seedlings in high-oil maize.

[0009] To achieve the above objectives, the present invention provides a rapid detection method for the number of seedlings in high-oil maize, comprising the following steps:

[0010] S1. Select high-oil corn monitoring plots;

[0011] S2. Acquire multi-dimensional field data for high-oil corn monitoring by using drones equipped with image acquisition devices;

[0012] S3, Multi-dimensional data preprocessing;

[0013] S4. Seedling segmentation and identification;

[0014] S5. Plant count statistics and verification;

[0015] S6. Data storage and analysis.

[0016] Preferably, in S1, high-oil maize monitoring plots are selected, specifically representative plots with different soil types, planting densities of 35,000-65,000 plants / hectare, and topographical conditions, as monitoring plots.

[0017] Preferably, in S2, an image acquisition device is used on a drone to acquire multi-dimensional data from the field for monitoring high-oil corn, as detailed below:

[0018] S21. Data collection for high-oil corn monitoring fields is carried out using drones equipped with RGB cameras and satellite positioning systems.

[0019] S22. The acquired multi-dimensional image data of high-oil corn monitoring fields are as follows:

[0020] (1) Spatial data: Orthophotos of the monitored fields were acquired by drones with a resolution of ≤2cm / pixel;

[0021] (2) Spectral data: Collect the RGB channel reflectance of the seedling canopy in the monitoring field;

[0022] (3) Structural data: Based on the structure of motion recovery (SfM) method, a digital surface model (DSM) containing the seedling projected leaf area and plant height point cloud was constructed.

[0023] Preferably, in S3, multi-dimensional data preprocessing is performed, the specific details of which are as follows:

[0024] S31. Noise Reduction Processing: Eliminate flight vibration noise through Non-Local Means filtering algorithm;

[0025] S32. Light correction: Compensating for leaf reflection based on the Retinex theoretical model;

[0026] S33. Contrast Enhancement: The CLAHE algorithm, which limits contrast adaptive histogram equalization, is used to improve the separation between seedlings and the background.

[0027] Preferably, in S4, seedling segmentation and identification are performed, with the following specific details:

[0028] S41. Use the dual-branch convolutional network DC-Net to segment seedlings and generate a binarized segmentation mask;

[0029] S42. Supplement the missing targets in the segmentation mask using the optimized YOLOv10 model;

[0030] S43, combined with the results of S41 and S42, performs morphological opening operations to eliminate adhesion; and uses a connected component labeling algorithm to label independent individual plants.

[0031] Preferably, in S41, seedling segmentation is performed using a dual-branch convolutional network DC-Net to generate a binarized segmentation mask, as detailed below:

[0032] (1) Branch 1 extracts the seedling pixel-level mask through the U-Net architecture;

[0033] (2) Branch 2 uses the color model HSV and the exponent ExG to generate a prior segmentation map;

[0034] (3) Combine the output results of the two branches to generate a binary segmentation mask; where seedling region = 1 and background = 0.

[0035] Preferably, in S42, the optimized YOLOv10 model is used to supplement the missing targets in the segmentation mask. The specific optimization is as follows:

[0036] (1) Replace the YOLOv10 backbone network with GhostNetV2 to adapt to embedded deployment;

[0037] (2) Add a channel-space dual attention reflective sensing module (GAM) to the end of the backbone to suppress reflective interference from high-oil corn leaves;

[0038] (3) Introduce a loss function that combines the enhanced cross-union ratio loss function EIoU Loss and the focal loss function FocalLoss to reduce the false negative rate of dense seedlings;

[0039] Preferably, in S5, the plant count and verification are performed, and the specific details are as follows:

[0040] S51. Automatic counting: Generates a plant count chart based on the centroid coordinates of connected components to count the number of individual plants in the monitored field.

[0041] S52, Accuracy Verification:

[0042] (1) Manual verification was conducted by randomly selecting 5% of the area;

[0043] (2) Calculate the F1-score to evaluate the counting accuracy, as shown below:

[0044]

[0045] Precision is the accuracy rate; Recall is the recall rate.

[0046] Preferably, in S6, data storage and analysis are performed as follows:

[0047] S61. Integrate and store multi-dimensional data generated by S1-S5;

[0048] S62. Conduct agronomic analysis based on stored data;

[0049] First, calculate the density per acre, as shown below:

[0050]

[0051] Secondly, the seedling uniformity index is calculated, as shown below:

[0052]

[0053] Wherein, ECI is the seedling uniformity index; N_actual is the actual number of seedlings that have emerged; N_expected is the expected number of seedlings that have emerged; and CV_height is the seedling height variation coefficient.

[0054] A rapid detection system for the number of seedlings in high-oil maize includes an image acquisition module, an image preprocessing module, a target segmentation module, a counting and statistics module, and a data storage and analysis module.

[0055] The image acquisition module includes a drone, an RGB camera, a satellite positioning system, and an image transmission device; this module is used to acquire image data of the monitored field and transmit the data to the image preprocessing module through the image transmission device.

[0056] The image preprocessing module includes an embedded GPU that sequentially runs the Retinex algorithm and the CLAHE algorithm to form an algorithm chain, eliminates environmental interference, enhances target features, and then inputs the processed data into the target segmentation module.

[0057] The target segmentation module includes a dual-branch convolutional network DC-Net model and an optimized YOLOv10 model, used to separate seedlings from the background, and the segmentation results are fed into the counting and statistics module.

[0058] The counting and statistics module outputs the plant count detection results and spatial distribution through connected component analysis;

[0059] The data storage and analysis module stores the detection results from the counting and statistics module and provides data analysis functions to help users understand the corn growth status.

[0060] The image preprocessing module, target segmentation module, and counting and statistics module run on an embedded GPU. The model is accelerated by FP16 quantization using the TensorRT tool, adapted for mobile deployment in the field, and generates field plant number heatmaps and early warning coordinates of missing seedling areas in real time.

[0061] Therefore, the present invention employs the above-mentioned rapid detection method and system for the number of seedlings in high-oil maize, which has the following beneficial effects:

[0062] (1) The method and system of the present invention can quickly and accurately detect the number of seedlings of high-oil corn, thereby improving the efficiency and accuracy of field management.

[0063] (2) The method and system of the present invention have wide applicability and can be used for field management of maize at different growth stages and in different growth environments.

[0064] (3) The method and system of the present invention provide data support for the analysis and prediction of maize growth status, which helps to optimize field management measures and improve maize yield and quality.

[0065] (4) The present invention utilizes image processing technology to achieve automatic recognition and counting, avoiding the inaccuracy and time-consuming nature of manual counting.

[0066] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0067] Figure 1 This is a technical roadmap for a rapid detection method for the number of seedlings in high-oil maize.

[0068] Figure 2 This is a structural diagram of a rapid detection system for the number of seedlings in high-oil corn. Detailed Implementation

[0069] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0070] like Figure 1 As shown, a rapid detection method for the number of seedlings in high-oil maize includes the following steps:

[0071] S1. Select high-oil corn monitoring plots;

[0072] S2. Acquire multi-dimensional field data for high-oil corn monitoring by using drones equipped with image acquisition devices;

[0073] S3. Preprocessing multi-dimensional data;

[0074] S4. Seedling segmentation and identification;

[0075] S5. Plant count statistics and verification;

[0076] S6. Data storage and analysis.

[0077] Example

[0078] A rapid detection method for the number of seedlings in high-oil maize includes the following steps:

[0079] S1. Select high-oil corn monitoring plots.

[0080] Representative fields with different soil types, planting densities of 35,000-65,000 plants / hectare, and different terrain conditions were selected as monitoring fields to ensure the model's generalization ability.

[0081] S2. Using drones equipped with image acquisition devices, acquire multi-dimensional data from high-oil corn monitoring fields, as detailed below:

[0082] S21. Data collection for high-oil corn monitoring fields using RGB drones, details of which are as follows:

[0083] The drone model is DJI M300 RTK, equipped with a 20-megapixel RGB camera and PPK positioning system.

[0084] Flight conditions are as follows:

[0085] (1) Choose a flight time that is clear and cloudless, with light intensity >80000 lux and solar altitude angle >30° to reduce shadow interference.

[0086] (2) Select corn seedlings that are in the 4-8 leaf stage, with a height of 15-40cm and a clear canopy structure.

[0087] (3) The flight altitude is 30m, the ground resolution is 1.2cm, the forward overlap is 80%, and the lateral overlap is 70%.

[0088] S22. Collect over 10,000 multi-dimensional images of high-oil corn monitoring fields, covering sunny and cloudy conditions, weeds, and stubble scenes. Use the Labelme tool to label the "minimum bounding rectangle of seedlings." The specific content of the multi-dimensional image data is as follows:

[0089] (1) Spatial data: Orthophotos of the monitored fields were acquired by drones with a resolution of ≤2cm / pixel.

[0090] (2) Spectral data: Collect the reflectance of the R, G, and B channels of the seedling canopy in the monitoring field.

[0091] (3) Structural data: Based on the structure of motion recovery (SfM) method, a digital surface model (DSM) containing the seedling projected leaf area and plant height point cloud was constructed.

[0092] S3. Preprocessing of multi-dimensional data, the specific steps are as follows:

[0093] S31. Noise Reduction Processing: Flight vibration noise is eliminated through a non-local means filtering algorithm.

[0094] S32. Light Correction: Based on the Retinex theoretical model, this method compensates for leaf reflection and specifically addresses the reflection problem in high-oil corn.

[0095] S33. Contrast Enhancement: The CLAHE algorithm, which limits contrast adaptive histogram equalization, is used to improve the separation between seedlings and the background.

[0096] S4. Seedling segmentation and identification.

[0097] S41. Using a dual-branch convolutional network (DC-Net), seedling segmentation is performed to generate a binary segmentation mask. The specific details are as follows:

[0098] (1) Branch 1 extracts the seedling pixel-level mask through the U-Net architecture.

[0099] (2) Branch 2 uses the color model HSV and the exponent ExG to generate a prior segmentation map.

[0100] (3) The output results of the two branches are merged to generate a binary segmentation mask, where seedling region = 1 and background = 0.

[0101] S42. Using the optimized YOLOv10 model, missing targets in the segmentation mask are supplemented. The specific optimizations are as follows:

[0102] (1) Replace the YOLOv10 backbone network with GhostNetV2, reducing the number of parameters by 58% and adapting to embedded deployment.

[0103] (2) Add a channel-space dual attention reflective sensing module (GAM) to the end of the backbone to suppress reflective interference from high-oil corn leaves.

[0104] (3) Introduce a loss function that combines the enhanced cross-union ratio loss function EIoU Loss and the focal loss function Focal Loss to reduce the false negative rate of dense seedlings.

[0105] S43, combined with the results of S41 and S42, performs morphological opening operations to eliminate adhesion, and labels independent individual plants using a connected component labeling algorithm.

[0106] S5. Plant count statistics and verification.

[0107] S51. Automatic counting: Generates a plant count chart based on the centroid coordinates of connected domains to count the number of individual plants in the monitored field.

[0108] S52, Accuracy Verification:

[0109] (1) Manual verification was conducted by randomly selecting 5% of the area.

[0110] (2) Calculate the F1-score to assess the counting accuracy. The requirement is that F1 ≥ 0.92 in complex field scenarios, as shown below:

[0111]

[0112] Precision is the accuracy rate; Recall is the recall rate.

[0113] S6. Data storage and analysis.

[0114] S61 integrates and stores the multi-dimensional data generated by S1-S5.

[0115] S62. Based on the stored data, perform agronomic analysis as follows:

[0116] The density per acre is calculated as follows:

[0117]

[0118] The seedling uniformity index was calculated as follows:

[0119]

[0120] Wherein, ECI is the seedling uniformity index; N_actual is the actual number of seedlings that have emerged; N_expected is the expected number of seedlings that have emerged; and CV_height is the seedling height variation coefficient.

[0121] like Figure 2 As shown, a rapid detection system for the number of seedlings in high-oil maize includes an image acquisition module, an image preprocessing module, a target segmentation module, a counting and statistics module, and a data storage and analysis module.

[0122] (1) The image acquisition module includes a DJI M300 RTK drone equipped with a 20-megapixel RGB camera, a PPK positioning system and an image transmission device, used to collect image data of the monitored fields and transmit the data to the image preprocessing module through the image transmission device.

[0123] (2) The image preprocessing module includes an NVIDIA Jetson AGX Orin embedded GPU that runs the Retinex+CLAHE algorithm chain to eliminate environmental interference, enhance target features, and transmit the processed data to the target segmentation module.

[0124] (3) The target segmentation module includes a dual-branch convolutional network DC-Net model and an optimized YOLOv10 model, which accurately separate seedlings from the background; among them, the DC-Net model has 4.8M parameters and an inference speed of 23fps@1080p.

[0125] (4) Counting and statistics module: Based on the output results of the target segmentation module, the number of plants and spatial distribution are output through connected component analysis.

[0126] (5) Data storage and analysis module: Stores the detection results of the counting and statistics module and provides data analysis functions to help users understand the growth status of corn.

[0127] (6) The image preprocessing module, target segmentation module, and counting and statistics module run on the embedded GPU. The TensorRT tool is used to accelerate the model with FP16 quantization, adapt to field mobile deployment, and generate field plant number heat map and early warning coordinates of seedling-deficient areas in real time.

[0128] Therefore, the present invention employs the aforementioned rapid detection method and system for the number of seedlings in high-oil maize, utilizing image processing technology to achieve automatic identification and counting, avoiding the inaccuracy and time-consuming nature of manual counting. This enables rapid and accurate detection of the number of seedlings in high-oil maize, improving the efficiency and accuracy of field management. The method and system of the present invention provide data support for the analysis and prediction of maize growth status, helping to optimize field management measures, improve maize yield and quality, and can be used for maize field management in different growth stages and environments, exhibiting wide applicability.

[0129] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A rapid detection method for high oil corn stand count at the seedling stage, characterized in that, The method comprises the following steps: S1, selecting a high-oil corn monitoring field; S2, acquiring multi-dimensional data of the high-oil corn monitoring field by a UAV carrying an image acquisition device; S3, multi-dimensional data preprocessing; S4, seedling segmentation and identification, the specific content being as follows: S41, performing seedling segmentation by a double-branch convolutional network DC-Net to generate a binary segmentation mask, the specific content being as follows: (1) Branch 1 extracts a seedling pixel-level mask through a U-Net architecture; (2) Branch 2 fuses a prior segmentation map generated by a color model HSV and an index ExG; (3) The binary segmentation mask is generated by fusing the output results of the two branches, wherein the seedling area is 1 and the background is 0; S42, supplementing the missed targets of the segmentation mask by a YOLOv10 model optimized as shown below: (1) Replacing the YOLOv10 backbone network with a GhostNetV2 to adapt to embedded deployment; (2) Adding a channel-spatial dual attention back reflection perception module GAM at the end of the Backbone to suppress the interference of high-oil corn leaf reflection; (3) Introducing an enhanced intersection over union loss function EIoU Loss and a focal loss function Focal Loss to reduce the dense seedling miss detection rate; S43, performing morphological opening operation on the results of S41 and S42 to eliminate adhesion, and labeling independent single plants by a connected domain labeling algorithm; S5, plant number statistics and verification; S6, data storage and analysis.

2. The method for rapid detection of high oil corn seedling plant number according to claim 1, characterized in that, In S1, a high-oil corn monitoring field is selected, and the specific selection is as follows: representative field blocks covering different soil types, a planting density of 35-65 thousand plants per hectare, and terrain conditions are selected as the monitoring field blocks.

3. The method for rapid detection of high oil corn seedling plant number according to claim 1, characterized in that, In S2, multi-dimensional data of the high-oil corn monitoring field is acquired by a UAV carrying an image acquisition device, and the specific content is as follows: S21, a UAV carrying an RGB camera and a satellite positioning system is used to collect data of the high-oil corn monitoring field; S22, the acquired multi-dimensional image data of the high-oil corn monitoring field is as follows: (1) Spatial data: orthographic images of the monitoring field are acquired by a UAV, and the resolution is less than or equal to 2 cm / pixel; (2) Spectral data: the seedling canopy RGB channel reflectivity of the monitoring field is collected; (3) Structural data: a digital surface model DSM containing seedling projection leaf area and plant height point cloud is constructed based on a structure from motion SfM method.

4. The method for rapid detection of high oil corn seedling plant number according to claim 1, characterized in that, In S3, multi-dimensional data preprocessing, the specific content is as follows: S31, denoising: Non-Local Means algorithm is used to eliminate flight vibration noise; S32, illumination correction: leaf reflection is compensated based on a Retinex theory model; S33, contrast enhancement: CLAHE algorithm is used to improve the separation degree between seedlings and background.

5. The method for rapid detection of high oil corn seedling plant number according to claim 1, characterized in that, In S5, plant number statistics and verification, the specific content is as follows: S51, automatic counting: the number of plants in the monitoring field is counted based on the connected domain centroid coordinates to generate a plant number statistics chart; S52, accuracy verification: (1) Randomly extract 5% of the area for manual review; (2) Calculate the F1-score to evaluate the counting accuracy, as follows: ； Where, Precision is the precision; Recall is the recall.

6. The method for rapid detection of high oil corn seedling plant number according to claim 1, characterized in that, In S6, data storage and analysis, the specific content is as follows: S61, integrate the multi-dimensional data generated by S1-S5; S62, based on the stored data, perform agronomic analysis; First, calculate the density per mu, as follows: ； Second, calculate the uniformity index, as follows: ； Where, ECI is the emergence uniformity index; N_actual is the actual emergence number; N_expected is the expected emergence number; CV_height is the coefficient of variation in height.

7. The method for rapid detection of high-oil maize seedling plant number according to claims 1-6, applied to a rapid detection system for high-oil maize seedling plant number, characterized in that, The system includes an image acquisition module, an image preprocessing module, a target segmentation module, a counting and statistical module, a data storage and analysis module; The image acquisition module includes a UAV, an RGB camera, a satellite positioning system, and an image transmission device; This module is used to collect image data of the monitored field, and transmits the data into the image preprocessing module through the image transmission device; The image preprocessing module includes an embedded GPU, Retinex algorithm and CLAHE algorithm running in sequence to form an algorithm chain, which eliminates environmental interference and enhances target features, and transmits the processed data into the target segmentation module; The target segmentation module includes a double-branch convolutional network DC-Net model and an optimized YOLOv10 model, which is used to separate seedlings from the background, and transmits the segmentation results to the counting and statistical module; The counting and statistical module outputs the plant number detection results and spatial distribution through connected component analysis; The data storage and analysis module stores the detection results of the counting and statistical module and provides data analysis functions to help users understand the growth status of corn; The image preprocessing module, target segmentation module and counting and statistical module run on the embedded GPU, and the model is quantized and accelerated by FP16 through the TensorRT tool, which is suitable for field mobile deployment and real-time generation of field plant number heat map and seedling shortage area warning coordinates.

Images