A seed meter seed meter performance detection method and device based on high frame line scanning

Abstract

The present application relates to a kind of based on high frame line scanning's seed metering device performance detection method and device, it is suitable for the performance detection of seed metering device.The detection method of the present application is to use linear array camera to the falling state of seed flow in the process of seed metering real-time image information collection, after image transmission to image processing system, information processing analysis is carried out, information processing analysis includes through image pre-processing and based on the training of improved deep learning network model accurately obtains the time sequence spatial distance of seed flow in seed metering, analyzes and calculates to obtain qualified index, miss index and rebroadcast index such as seed metering device key performance indicators, realizes the seed metering performance detection of seed metering device;The device applied in the present application includes rack, seed metering device, image acquisition device, image processing device, inoculation dish, integrated aluminum frame.The present application provides scientific basis for the accuracy and high efficiency research and development of the core component of seeding machine agricultural machinery, seed metering device.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of seed metering performance detection, in particular to a seed metering performance detection method and device based on high-frame line scanning. BACKGROUND

[0002] The application discloses a precision seed metering performance detection method and device based on high-frame line scanning, and is suitable for performance detection of a seed metering device. The detection method is to use a linear array camera to collect real-time images of falling states of seed flow in a seed metering process, and transmit the images to an image processing system for information processing and analysis. The information processing and analysis includes accurately obtaining time-space distances of the seed flow in the seed metering process through image preprocessing and a modified deep learning network model training, analyzing and calculating key performance indicators of the seed metering device such as a qualified index, a missing planting index and a re-planting index, and realizing seed metering performance detection of the seed metering device. The device applied in the application comprises a rack, a seed metering device, an image collection device, an image processing device, a seed planting disc and an integrated aluminum frame. The application provides a scientific basis for the accuracy and efficiency research of the seed metering device which is a core component of a seeding machine. SUMMARY

[0003] In view of the above problems, the present application provides a seed metering performance detection method and device based on high-frame line scanning.

[0004] The application adopts the following technical scheme:

[0005] A precision seed metering performance online detection method based on high-frame line scanning, comprising the following steps:

[0006] Step 1, install the seed metering device to be detected on the rack, set the required working parameters, start the seed metering device, and the linear array camera continuously shoots images of a horizontal linear field of view in front according to a preset frame rate, and continuously collects time sequence image information of seeds in the seed flow discharged by the seed metering device when the seeds pass through the linear field of view;

[0007] Step 2, pre-process the effective frames of the collected time sequence images; input the images into a trained detection model, splice a spatial distribution state image of the seed flow after the detection model predicts the pre-processed images, output relative position information and a total number of seeds of various seeds in the seed flow in the process of falling into the linear field of view according to the frame rate of the linear array camera and the shooting time of each image;

[0008] Step 3, according to the relative position information of various seeds obtained in step 2, obtain the vertical spacing of adjacent seeds in the seed flow in the process of falling into the linear field of view, and let the vertical spacing of the adjacent i th seed and the i+1 th seed be ΔL i,i+1 , then the actual seed metering time interval of the i th seed and the i+1 th seed is Wherein v2 is the speed of the seed falling into the horizontal linear field of view of the linear array camera, which is a fixed value;

[0009] Step 4, calculate the theoretical seed spacing interval Δt, compare the theoretical seed spacing interval Δt with the actual seed spacing interval Δt i,i+1 , judge whether the seed spacing of the i+1th seed is qualified, missed or repeated, after judging all the seeds in a batch of seed flow, divide the qualified number, the missed number and the repeated number by the total number of seeds respectively to calculate the qualified index, the missed index and the repeated index of the seed spacing device.

[0010] Further, the step 4 specifically includes the following steps:

[0011] Step 4.1, calculate the theoretical seed spacing interval In the formula, n is the rotation speed of the seed spacing disc, r / min; z is the number of holes;

[0012] Step 4.2, calculate the actual seed spacing interval Δt of the seed spacing device 实 Compare the actual seed spacing interval with the theoretical seed spacing interval, if 0.5Δt 理 ≤Δt 实 ≤1.5Δt 理 , it is determined as normal seeding, Δt 实 >1.5Δt 理 , it is determined as missed seeding, Δt 实 <0.5Δt 理 , it is determined as repeated seeding, and the repeated number, the missed number and the actual seed spacing number are obtained;

[0013] Step 4.3, the qualified index = the number of seeds of normal seeding / total number of seeds, the missed index = the missed number / total number of seeds, the repeated number = the repeated number / total number of seeds, and the qualified index, the missed index and the repeated index are calculated according to the seed spacing information of a batch of seed flow.

[0014] Further, in the step 3 In the formula, R is the radius of the seed spacing disc, n is the rotation speed of the seed spacing disc, y0 is the distance from the hole spacing seed spacing device outlet just entering the positive pressure area to the seed spacing device outlet, and y1 is the vertical distance from the seed spacing outlet to the lens of the linear array camera.

[0015] Further, the method of pre-processing the image in step 1 includes the following steps:

[0016] Step 1.1, convert the obtained seed image into a gray scale image;

[0017] Step 1.2, crop the left and right sides of the gray scale image to reduce the image size;

[0018] Step 1.3, analyze the histogram features of the cropped gray image, calculate the first-order forward difference of the pixel point number in the range of 0-10 gray levels on the original gray image histogram, take the gray value of the first-order difference maximum point as the gray segmentation threshold, and binarize and segment the cropped original gray image.

[0019] Step 1.4, using a circular structural element to perform morphological close-open smoothing processing on the binarized image, filter edge noise, and finally obtain the preprocessed image.

[0020] Further, the method for obtaining the trained detection model comprises the following steps:

[0021] Step 2.1, start the seed metering device, and the linear array camera continuously captures images of the front horizontal linear field of view at a preset frame rate, continuously collects time sequence image information of the seeds in the seed flow discharged by the seed metering device when passing through the linear field of view; filter the collected image effective frames, pre-process the grain image to obtain a pre-processed image, label and classify the detection targets in the pre-processed image to obtain a data set, the detection targets being single grains and adhered grains, and the data set being classified as one, two, and three, indicating that single grains, two-grain adhered grains, and three-grain adhered grains can be recognized, and the data set is divided into a training set and a validation set in proportion;

[0022] Step 2.2, training the deep learning network model using the training set, the model being used for target detection and outputting a detection result, the detection result including grain anchor frames, grain numbers, and position information, and the optimal model being obtained by minimizing a loss function;

[0023] Step 2.3, using the optimal model to perform target detection on the validation set to obtain a detection result, and determining whether the error between the detection result and the actual result meets a preset requirement, if yes, taking the optimal model as the final detection model, and if not, returning to step 2.2 to retrain the model.

[0024] Further, the content labeled in step 1 includes the coordinates of the seed and the seed bounding box and seed category label information, and a corresponding txt file is generated for the labeled content.

[0025] Further, the pre-processed image is an image meeting the input requirements of the deep learning network base model.

[0026] A device for detecting the seed metering performance of a seed metering device based on high-frame line scanning, comprising a rack, an image acquisition device, an image processing device, a seeding disc, an integrated aluminum frame, and a seed metering device to be detected, which is detachably arranged on the rack, and the image acquisition device and the image processing device are arranged on the integrated aluminum frame.

[0027] The seed metering device to be detected is used to vertically seed the linear array camera below.

[0028] The image acquisition device is used for acquiring time sequence image information of the seeds in the seed flow discharged by the seed metering device when passing through the linear field of view, and sending the image information to the image processing system.

[0029] The image processing system is used for receiving and processing and analyzing the grain image information acquired by the linear array camera, and displaying the processing result.

[0030] Further, the image acquisition device comprises a camera module, a light source module and an integrated aluminum frame.

[0031] The camera module comprises an acquisition card, a camera support frame, a linear array camera, an adapter ring and a camera lens, the image acquisition card is connected with the linear array camera through a CameraLink interface, the camera lens and the linear array camera are connected through threads of the adapter ring, and the linear array camera is placed on two crossbeams of the integrated aluminum frame and is fixed in position by the camera support frames on the left and right sides.

[0032] The light source module comprises a strip-shaped light source, a light source support, an L-shaped adjustable handle and a light source controller, the strip-shaped light source is fastened on the light source support, the light source support is placed on the beam of the frame and can be rotated at a preset angle along the crossbeam.

[0033] The integrated aluminum frame comprises an aluminum profile and a connecting plate, the upper part of the integrated aluminum frame is composed of left and right top plates, the right top plate is fastened on the aluminum profile, the left top plate is connected with the right top plate through a hinge, one side is fixed and the other side is movable, and the left and right top plates are used for adjusting the irradiation angle of the light source module before image acquisition, a partition plate is fixed in the middle part of the integrated aluminum frame, the light source controller is placed at the bottom of the frame, and a ventilation opening is drilled in the lower part of the integrated aluminum frame.

[0034] Compared with the prior art, the above technical scheme has the following advantages:

[0035] 1. Since the theoretical height of the linear array camera for acquiring images along the motion direction is not limited, the monitoring of the grain target under high-speed dynamics can be realized, the position relationship between the seeds can be observed after conventional image shooting, the seed arrangement sequence of different seeds in the same image cannot be determined, the seed arrangement sequence and the relative position of the seeds can be accurately judged by adopting the method, the seed arrangement performance can be judged, and thus a new technical reference means and a preliminary scientific theoretical basis are provided for the seed arrangement performance detection.

[0036] 2. The system can not only detect the seed arrangement uniformity of the single-grain precision seed metering device in real time and on line on the basis of effectively monitoring the small-particle seed flow of rapeseed, but also has certain applicability for the seeding amount monitoring of the strip seeding device.

[0037] 3, Put forward a kind of real-time metering method for strip sowing seeder sowing amount using linear array camera, realize the detection of uniformity of sowing, can more clearly and accurately grasp the influence of sowing device structure on uniformity of sowing, provide key performance detection equipment for realizing precision seeding.

[0038] 4, Put forward a seed counting method based on YOLO series algorithm of deep learning, realize low cost, high accuracy of automatic counting of grain.

[0039] 5, Simple structure, easy to operate, convenient for performance detection of sowing device.

[0040] The application will be described in detail below in combination with the drawings and examples. DRAWINGS

[0041] Figure 1 It is the schematic diagram of the sowing performance detection device of the application based on high frame line scanning;

[0042] Figure 2 It is Figure 1 sowing device structure schematic diagram;

[0043] Figure 3 It is Figure 1 detection device structure schematic diagram;

[0044] Figure 4 It is Figure 3 camera module schematic diagram;

[0045] Figure 5 It is Figure 3 light source module schematic diagram;

[0046] Figure 6 It is the system workflow diagram of the application;

[0047] Figure 7 It is the principle diagram of the detection method of the application;

[0048] Figure 8 It is the visual training result diagram.

[0049] The component list represented by each number is as follows:

[0050] 1, rack; 2, seed metering device; 3, detection device; 4, human-computer interaction interface; 201, seed metering device positive pressure pipeline; 202, motor frame; 203, stepper motor; 204, elastic coupling; 205, seed metering device; 206, seed metering device negative pressure pipeline; 207, AC speed regulation fan; 301, integrated aluminum frame; 302, light source controller; 303, case; 304, partition; 305, camera module; 306, light source module; 307A, left top plate; 307B, right top plate; 308, hinge; 309, inoculation disc; 310, connecting plate; 301A, camera support frame; 301B, linear array camera; 301C, connecting ring; 301D, camera lens; 306A, L-shaped adjustable handle; 306B, light source support; 306C, strip-shaped light source. DETAILED DESCRIPTION

[0051] The principles and features of the present application are described below in conjunction with the accompanying drawings, and the examples are only used to explain the present application and are not used to limit the scope of the present application.

[0052] As shown in Figures 1-5 , a high-line scanning-based precision seed metering device seed metering performance detection method and device, comprising a rack 1, a seed metering device 2, an image acquisition device, an image processing device, an inoculation disc 309, and an integrated aluminum frame 301; the image acquisition device, the image processing device, the integrated aluminum frame, and the connecting plate together constitute a detection device 3.

[0053] In combination with Figure 1 , the rack is assembled from 4040 aluminum profiles, and is fixed in the middle by hinges.

[0054] In combination with Figure 2 , the seed metering device 2 comprises a seed metering device 205, a fan 207, an elastic coupling 204, a stepper motor 203, a positive pressure pipeline 201, and a negative pressure pipeline 206. The seed metering device 205 is a pneumatic seed metering device, the fan 207 is used to provide positive pressure and negative pressure to the seed metering device 205, and cooperates with the seed metering device 205 to perform seed metering, the elastic coupling 204 is assembled on the stepper motor 203 and connected with the rotating shaft of the seed metering device 205, and the seed metering device 2 is detachably arranged on the rack 1; the seed metering device 2 is used to supply seeds to the image acquisition device.

[0055] In combination with Figure 3The detection device 3 includes an image acquisition device, an image processing device, an integrated aluminum frame 301, a connecting plate 310, a top plate 307, and a partition 304. The image acquisition device includes a camera module 305, a light source module 306, and a light source controller 302. The image processing device includes a chassis 303 and a human-machine interface 4. The detection device 3 is mounted on the platform 1 via two lower aluminum profiles, and the grain drop height can be changed by altering the distance between the supporting aluminum profiles and the ground. The human-machine interface 4 is fitted into the rectangular cut of the connecting plate on the left side of the detection device 3 and is connected to the chassis via a connecting cable.

[0056] In one implementation, the rectangular hole on the upper right top plate 307B of the detection device 3 is directly opposite the seed outlet of the seed metering device 205. Combined with the field of view of the industrial camera 301B and the working distance of the lens 301D, a rectangular window of 190×100mm is cut to facilitate seed dropping and seed image capture. An inoculation tray 309 is provided at the bottom of the device for easy seed recovery and utilization. The device for detecting the performance of the seed metering device based on high-frame-rate line scanning is characterized in that the industrial camera 301B is a linear array scanning element. The linear array scanning element performs line-by-line scanning due to the relative motion between the camera and the object being detected. Through high-frequency scanning, images of each line of the object being detected are acquired, and these linear images are finally output. These images are then reassembled into a planar array image by stacking them according to the order of acquisition. By setting the scanning frequency of the industrial camera 301B, high-speed seed target detection can be successfully achieved.

[0057] In one implementation, the right top plate 307B of the detection device 3 has four through holes with a diameter of 6mm drilled above it, which can be fastened to the rectangular member by bolts and nuts. The left top plate 307A and the right top plate 307B are connected by a hinge 308, which allows the left top plate 307A to be opened and closed. Before image acquisition, the position and angle of the light source 306C can be adjusted by opening the left top plate 307A. During image acquisition, the top plate can be closed to isolate external environmental interference.

[0058] In one embodiment, the detection device 3 has an inoculation tray 309 at its bottom, which is located directly below the rectangular window of the right top plate 307B.

[0059] Combination Figure 4The camera module 305 includes a camera support frame 301A, an industrial camera 301B, an adapter ring 301C, and a camera lens 301D. The industrial camera 301B is a Hikvision MV-CL082-92CM model with a resolution of 8192×2 and a maximum line frequency of 100KHz. The camera lens 301D is an LS8003A model. The industrial camera 301B and the camera lens 301D are fixed together by a threaded connection via the adapter ring 301C. The camera support frame 301A is located on both sides of the industrial camera 301B and is fixed with screws. The camera support frame is the same height as the line scan camera and has two 5mm diameter through holes drilled on both the top and bottom sides. It is fastened to the middle of the two crossbeams of the detection device with bolts and nuts to restrict camera movement. The partition plate 304 is located in the middle of the detection device and is fixed to the central aluminum profile with T-bolts.

[0060] Combination Figure 5 The light source module 306 includes two symmetrical L-shaped adjustable handles 306A, a light source bracket 306B, a strip light source 306C, and a light source controller 302. The L-shaped adjustable handles 306A and bolts and nuts can fix the light source bracket 306B to the aluminum profiles on the left and right sides. Both aluminum profiles are drilled with six equally spaced threaded holes with a diameter of 6mm and a hole spacing of 15mm, which are used to adjust the horizontal position and angle of the strip light source 306C illuminating the seeds. A 100mm long through groove is drilled on the outer connecting plate of the aluminum profile, which cooperates with the aluminum profile to fix the light source bracket.

[0061] The bar light source 306C uses an ultra-bright bar light source QLH-22516. The bar light source 306C is fixed to the light source bracket 306B by bolts and nuts. The adjustable angle range of the bar light source 306C is 0-90°.

[0062] Specific implementation process:

[0063] 1) Set the required working parameters, start the seed metering device, the seed metering stepper motor and AC speed-regulating fan start working, collect image information of each seed as it passes through the detection device, filter the valid frames of the collected images, preprocess the seed images to obtain preprocessed images, use the Labelimg tool to label and classify the detection targets in the preprocessed images to obtain the dataset;

[0064] The detection targets are single grains and adhered grains;

[0065] The dataset is divided into a training set and a validation set in a ratio of 8:2.

[0066] 2) Based on the YOLOV7 network basic model, a BiFormer module is added, the Focal Loss loss function is introduced, Cutout is used on Mosaic, and the EfficientNet lightweight feature extraction network is used to compress the network configuration to obtain an improved YOLOV7 network model.

[0067] 3) Train and iterate the improved YOLOv7 network model using the training set to obtain the optimal model; use the optimized model to perform target detection on the validation set to obtain the detection results, which include seed anchor boxes and seed number and location information.

[0068] 4) Combining Figure 7 Based on the number of seeds and the parameter information between seeds in each grayscale image of the seed stream obtained in step 3, the performance parameters of the seed metering device are calculated: by calculating the vertical velocity v1 of the seed leaving the seed drop outlet, as well as the rotational speed n and radius R of the seed metering disc, the instantaneous velocity v2 of the seed when it leaves the straight line segment y1 and is scanned by the line scan camera can be obtained. By calculating the instantaneous velocity v2 of the seed when it enters the field of view of the industrial camera and the vertical physical distance L between adjacent seeds obtained from the seed position information in the image output by the line scan camera, the actual seed metering time interval can be obtained. The specific calculation formula is as follows:

[0069]

[0070] In the formula, n is the rotational speed of the seed metering disc, r / min; z is the number of holes.

[0071] Calculate the actual seeding time interval Δt when adjacent seeds leave the seed guide and enter the detection device in the seed stream. 实 In the formula, L is the actual physical distance between the centroids of adjacent seeds captured in the output image when they pass through the field of view of the line array camera, v1 is the vertical velocity of the seed leaving the seed guide, v2 is the seed velocity captured by the line array camera when the seed falls a length y1 from the seed guide, y0 is the distance between the hole and the seed metering device outlet, and y1 is the vertical distance between the seed drop outlet and the camera lens.

[0072] The seed flow performance indicators after leaving the seed metering device are obtained by comparing with the theoretical seeding time interval. The seeding performance indicators include the qualified index, the missed seeding index, and the reseeding index.

[0073] Combination Figure 8 'a' represents the first seed that passes through the linear field of view of the linear array camera, i.e., the first seed captured by the camera. The next seed, 'b', passes through Δt... 实 The time is captured by the camera. The vertical distance L1 between the two seeds in the output image is the number of pixels in the vertical direction of the adjacent seed centers multiplied by the physical size of each pixel. It is also the distance Δt traveled by the two seeds at a velocity v2.实 The time is the distance traveled by the seeds scanned sequentially. Within a certain time frame, seeds c, d, and e pass through the linear field of view of the line scan camera in sequence and are successfully captured. In the image, the colored background next to the seeds represents the model's classification and confidence level of the prediction results. The confidence level is usually represented by a value between 0 and 1, i.e., the probability of seed a being "One" is 0.91, the probability of seed b being "Two" is 0.82, the probability of seed c being "One" is 0.90, the probability of seed d being "One" is 0.94, and the probability of seed e being "One" is 0.86. The actual seeding time interval Δt between two adjacent seeds is calculated by the number of pixels in the vertical direction of two adjacent seeds in the image and the instantaneous velocity of the seeds at the moment of scanning by the line scan camera. 实 By comparing the actual physical distance between adjacent grains in the vertical direction, it can be determined whether there is any double sowing or missed sowing between adjacent grains.

[0074] As one implementation method, step 4, the method for calculating the pass index, the missed broadcast index, and the rebroadcast index, includes the following steps:

[0075] 2.1) Calculate the theoretical time interval for ovulation. In the formula, n is the rotational speed of the seed metering disc, r / min; z is the number of holes.

[0076] 2.2) Calculate the actual seeding time interval Δt between adjacent seeds leaving the seed guide and entering the field of view of the linear array camera in the seed stream. 实 The actual seeding interval is compared with the theoretical seeding interval. If 0.5Δt 理 ≤Δt 实 ≤1.5Δt 理 Then it is determined to be normal sowing, Δt 实 >1.5Δt 理 Then it is determined to be a missed broadcast, Δt 实 <0.5Δt 理 It is then judged as a replay;

[0077] 2.3) The experiment used the seed metering device's seed drop port and the line scan camera below the seed drop port to detect the time data of the seed metering process, collect images of the spatial distribution of the seed flow, continuously monitor the seed metering time interval of 251 seeds, and output the data of this batch of seed flow to calculate key performance indicators of the seed metering device such as the qualification index, the missed seeding index, and the reseeding index.

[0078] The above description provides examples of the preferred embodiments of the present invention. Parts not detailed herein are common knowledge to those skilled in the art. The scope of protection of the present invention is determined by the claims. Any equivalent modifications based on the technical teachings of the present invention are also within the scope of protection of the present invention.

Claims

1. A method for online detection of seed metering performance based on high frame line scanning, characterized in that, Includes the following steps: Step 1: Install the seed metering device to be tested onto the stand, set the required working parameters, start the seed metering device, and the line scan camera continuously captures images of the horizontal linear field of view in front of it at the preset frame rate. The time sequence image information of the seeds in the seed stream discharged by the seed metering device when they pass through the linear field of view is continuously collected. Step 2: Preprocess the acquired time-series image frames; input the images into the trained detection model, and after the detection model predicts the preprocessed images, it stitches them together to obtain the spatial distribution state image of the seed stream. Based on the frame rate of the linear scan camera and the shooting time of each image, it outputs the relative position information of various seeds in the seed stream during the process of falling into the linear field of view and the total number of seeds. Step 3: Based on the relative position information of various seeds obtained in Step 2, obtain the vertical distance between adjacent seeds in the seed stream during the process of falling into the linear field of view. Let the vertical distance ΔL between the i-th seed and the (i+1)-th seed be... i,i+1 Then the actual time interval between the i-th seed and the (i+1)-th seed falling is... ,in The velocity of the grains falling into the horizontal linear field of view of the line scan camera is a fixed value. In step 3 In the formula, R is the radius of the seed metering disc, n is the rotational speed of the seed metering disc, y0 is the distance from the orifice just entering the positive pressure zone to the seed metering outlet, and y1 is the vertical distance from the seed metering outlet to the lens of the linear array camera. Step 4: Calculate the theoretical time interval for ovulation. Compare the theoretical ovulation time intervals Time interval with actual seeding To determine whether the (i+1)th seed is planted correctly, missed, or replanted, after judging all seeds in a batch of seed streams, the number of correct, missed, and replanted seeds are divided by the total number of seeds to calculate the correct index, missed index, and replanting index of the seed metering device.

2. The online detection method for seed metering performance based on high frame line scanning according to claim 1, characterized in that, Step 4 specifically includes the following steps: Step 4.1: Calculate the theoretical time interval for seeding. In the formula, n is the rotational speed of the seed metering disc, r / min; z is the number of holes. Step 4.2: Calculate the actual seeding interval of the seed metering device. The actual seeding interval is compared with the theoretical seeding interval. If... This is considered normal sowing. This is considered a missed broadcast. It is then determined to be a replay, and the number of replays, the number of missed replays, and the actual number of seedings are obtained; Step 4.3: Qualification index = number of seeds sown normally / total number of seeds; Missed sowing index = number of missed sowings / total number of seeds; Re-sowings = number of re-sowings / total number of seeds. The qualification index, missed sowing index and re-sowing index are calculated based on the seeding information of a batch of seeds.

3. The online detection method for seed metering performance based on high frame line scanning according to claim 1, characterized in that, The image preprocessing method in step 1 includes the following steps: Step 1.1: Convert the obtained seed image into a grayscale image; Step 1.2: Crop the grayscale image on both sides to reduce its size; Step 1.3: Analyze the histogram features of the cropped grayscale image, calculate the first-order forward difference of the number of pixels in the grayscale range of 0 to 10 on the histogram of the original grayscale image, and use the grayscale value of the maximum point of the first-order difference as the grayscale segmentation threshold to perform binarization segmentation on the cropped original grayscale image. Step 1.4: Use circular structuring elements to perform morphological smoothing (closing then opening) on ​​the binarized image to filter edge noise and finally obtain the preprocessed image.

4. The online detection method for seed metering performance based on high frame line scanning according to claim 1, characterized in that, The method for obtaining a trained detection model includes the following steps: Step 2.1: Start the seed metering device. The linear array camera continuously captures images of the horizontal linear field of view in front of the seed metering device at a preset frame rate. It continuously collects time-series image information of the seeds in the seed stream discharged by the seed metering device as they pass through the linear field of view. Filter the valid frames of the collected images, preprocess the seed images to obtain preprocessed images, and label and classify the detection targets in the preprocessed images to obtain a dataset. The detection targets are single seeds and adhering seeds. The dataset is classified as one, two, and three, indicating that it can identify single seeds, two adhering seeds, and three adhering seeds. The dataset is divided into training set and validation set according to the ratio. Step 2.2: Train the deep learning network model using the training set. The model is used for object detection and outputs detection results, including seed anchor boxes, seed number and their location information. The optimal model is obtained by minimizing the loss function. Step 2.3: Use the optimal model to perform target detection on the validation set, obtain the detection results, and determine whether the error between the detection results and the actual results meets the preset requirements. If yes, use the optimal model as the final detection model; otherwise, go to step 2.2 to retrain the model.

5. The online detection method for seed metering performance based on high frame line scanning according to claim 1, characterized in that, The annotations in step 1 include the coordinates of the seed and its bounding box, as well as the seed category label information. A corresponding txt file is generated for each annotation.

6. The online detection method for seed metering performance based on high frame line scanning according to claim 1, characterized in that, The preprocessed image is an image that meets the input requirements of the basic model of the deep learning network.

7. A device for detecting the seed metering performance of a seed metering device based on high frame line scanning, wherein the detection is performed based on the online detection method for seed metering performance based on high frame line scanning as described in claim 1, characterized in that, It includes a frame, a seed metering device, an image acquisition device, and an image processing device. The seed metering device to be tested can be detachably mounted on the frame, and the image acquisition device and the image processing device are mounted on an integrated aluminum frame. The seed metering device to be tested is used to seed vertically towards the downward-facing line scan camera; The image acquisition device is used to acquire time-series image information of seeds in the seed stream discharged by the seed metering device as they pass through the linear field of view, and to send the image information to the image processing device. The image processing device is used to receive and process the grain images acquired by the line scan camera, and display the processing results.

8. The device for detecting the seed metering performance of a seed metering device based on high frame line scanning according to claim 7, characterized in that, The image acquisition device includes a camera module, a light source module, and an integrated aluminum frame; The camera module includes an image acquisition card, a camera support frame, a line scan camera, an adapter ring, and a camera lens. The image acquisition card is connected to the line scan camera via a CameraLink interface. The camera lens and the line scan camera are connected by the threads of the adapter ring. The line scan camera is placed on the two crossbeams of the integrated aluminum frame and is fixed in position by the camera support frames on the left and right sides. The light source module includes a strip light source, a light source bracket, an L-shaped adjustable handle, and a light source controller; the strip light source is fixed to the light source bracket, which is placed on the beam of the integrated aluminum frame and can rotate at a preset angle along the beam; The integrated aluminum frame includes an aluminum profile and a connecting plate; the top of the integrated aluminum frame consists of two top plates, left and right, with the right top plate fastened to the aluminum profile and the left top plate connected to the right top plate by a hinge, one side being fixed and the other side being movable, used to adjust the illumination angle of the light source module before image acquisition; a partition is fixed in the middle of the integrated aluminum frame; the light source controller is placed at the bottom of the frame, and ventilation holes are drilled in the lower part of the integrated aluminum frame.

Images