Seedling precision sowing detection device and method

Abstract

The invention discloses a seedling precision sowing detection device and method. The device comprises supporting legs; the seedling tray conveying mechanism comprises a conveying support, a conveying belt, a conveying driving motor and a conveying control cabinet, the conveying support is installed on the supporting legs, the conveying belt, the conveying driving motor and the conveying control cabinet are installed on the conveying support, and the conveying motor is connected with the conveying belt and the conveying control cabinet; the detection box is mounted above the conveying bracket, and an image acquisition module is arranged in the detection box corresponding to the conveying belt; the detection control mechanism is mounted on the detection box, is connected with the conveying control cabinet, and is used for collecting images of the potted trays after seeding and analyzing and recording the seeding condition in the seedling seeding process, post-processing the collected images by adopting a deep learning model so as to obtain seeding information, and sending the seeding information to the conveying control cabinet. And storing the seeding information in a local database. The invention also provides a seedling precision sowing detection method.

Description

Technical Field

[0001] The invention relates to a precision seeding technology for seedling cultivation in a tray, and in particular to a detection device and method for precision seeding in a seedling cultivation. Background Art

[0002] Plug tray seedling cultivation technology is a modern seedling cultivation technology that uses light-based soilless materials such as peat and vermiculite as the seedling substrate, adopts mechanized precision sowing, and achieves one-time seedling formation. The seedling precision sowing production line is a representative equipment for plug tray seedling cultivation, which can realize automated and efficient sowing. Compared with manual sowing, it can greatly improve planting efficiency, save time and labor costs, reduce labor intensity, and improve production benefits. However, during the operation, manual real-time observation and statistics of the sowing operation are required, which is labor-intensive. Especially when operating at high speeds, it is difficult for the human eye to quickly distinguish seeds from the complex soil background, and the accuracy of judging the sowing status of seeds is not high. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a seedling precision sowing detection device and method in view of the above-mentioned defects of the prior art.

[0004] In order to achieve the above-mentioned object, the present invention provides a seedling precision sowing detection device, which includes:

[0005] Support leg;

[0006] The seedling tray conveying mechanism includes a conveying bracket, a conveyor belt, a conveying drive motor and a conveying control cabinet. The conveying bracket is installed on the support leg, and the conveyor belt, the conveying drive motor and the conveying control cabinet are respectively installed on the conveying bracket. The conveying motor is connected to the conveyor belt and the conveying control cabinet respectively.

[0007] A detection box is installed above the conveying bracket, and an image acquisition module is provided in the detection box corresponding to the conveyor belt; and

[0008] The detection control mechanism is installed on the detection box and connected to the conveying control cabinet. It is used to collect images of the hole tray after sowing and analyze and record the sowing conditions during the seedling sowing process. A deep learning model is used to post-process the collected images to obtain sowing information, and the sowing information is stored in a local database.

[0009] The above-mentioned seedling precision sowing detection device, wherein the detection control mechanism includes a speed adjustment unit, a light intensity adjustment unit, an image acquisition unit, a camera posture control unit, an image processing unit, a human-computer interaction unit, a data analysis unit, a data storage unit, a data transmission unit and a serial port expansion unit, and the serial port expansion unit is used for communication interaction between different devices; the data storage unit stores the detection default setting parameters, and the data analysis unit adopts a target detection algorithm to realize the recognition and detection of crops, and automatically matches the image information obtained by the sowing seed analysis.

[0010] In the above-mentioned seedling precision sowing detection device, the support legs are of a liftable structure to adjust the height of the conveyor belt to be consistent with the level of the conveyor belt of the seedling sowing production line.

[0011] The above-mentioned seedling precision sowing detection device, wherein the seedling tray conveying mechanism also includes a limit rod, which is installed on the conveying frame on both sides of the conveyor belt. The limit rod is used to constrain the seedling tray from lateral position deviation and twisting, and to ensure that the seedling tray posture is consistent during the forward process so that the collected seedling tray images are consistent; seedling trays of different specifications are constrained by adjusting the spacing of the limit rods.

[0012] The above-mentioned seedling precision sowing detection device also includes a seedling tray position detection mechanism, which is installed on the limit rod and connected to the detection control mechanism, forming an adjustable angle of 30° to 60° with the forward direction of the conveyor belt, and is used to detect whether the seedling tray is transported in place and trigger the picture acquisition instruction.

[0013] The above-mentioned seedling precision sowing detection device, wherein the detection box includes a box frame and a light shielding plate covered on the box frame; the image acquisition module includes a camera, and the detection box is also provided with a camera adjustment mechanism and a strip light source, the camera adjustment mechanism is installed at the upper end of the box frame of the detection box, and the camera is installed on the camera adjustment mechanism to adjust the shooting position and angle; the installation direction of the strip light source is parallel to the forward direction of the conveyor belt, and is tilted downward to illuminate the conveyor belt.

[0014] The above-mentioned seedling precision sowing detection device, wherein the camera adjustment mechanism includes a normal angle adjustment servo, a horizontal angle adjustment servo, a front and rear adjustment slide and a height adjustment slide, the front and rear adjustment slide includes a front and rear adjustment drive motor, a front and rear adjustment slide rail and a front and rear adjustment slider, the front and rear adjustment drive motor is connected to the front and rear adjustment slide rail, and the front and rear adjustment slider is installed on the front and rear adjustment slide rail; the height adjustment slide includes a height adjustment drive motor, a height adjustment slide rail and a height adjustment slider, the height adjustment drive motor is connected to the height adjustment slide rail, and the height adjustment slider is installed on the height adjustment slide rail; the front and rear adjustment slide rail is connected to the height adjustment slider; the height adjustment slide rail is connected to the box frame of the detection box through the height adjustment fixing piece; the camera is connected to the front and rear adjustment slider through a two-degree-of-freedom pan-tilt composed of the normal angle adjustment servo and the horizontal angle adjustment servo.

[0015] The above-mentioned seedling precision seeding detection device, wherein the normal angle adjustment servo adjusts the shooting angle of the camera to 0 to 120° in the normal plane of the conveyor belt conveying plane; the horizontal angle adjustment servo adjusts the shooting angle of the camera to 0 to 360° in the parallel plane of the conveyor belt conveying plane.

[0016] In order to better achieve the above-mentioned object, the present invention also provides a seedling precision sowing detection method, wherein the seedling precision sowing detection device is used to collect and analyze the hole tray sowing situation between the sowing and covering links of the seedling sowing production line, comprising the following steps:

[0017] S100, initializing detection parameters, loading the most recently used detection parameters stored in the local database or loading the default detection parameters, and resetting the detection parameters through the human-computer interaction interface;

[0018] S200, image acquisition: During the seedling raising and sowing process, the seedling tray position detection mechanism triggers the camera to capture images of the seedling tray after sowing;

[0019] S300, using a deep learning model to post-process the collected image to obtain seeding information; and

[0020] S400: Storing the sowing information in a local database, the sowing information including a sowing single seed rate, a reseeding rate, a missed seeding rate, and position coordinates of missed seeding cells.

[0021] In the above-mentioned seedling precision sowing detection method, in step S300, the deep learning model is a target detection algorithm model trained for the collected data sets of different seeds, and the processing after image collection further includes:

[0022] S301, after grayscale preprocessing, image morphological noise reduction, and Hough transform, the collected image is cropped according to the grid division to obtain the image and position coordinates of a single grid;

[0023] S302, calling the target detection algorithm model to perform identification and obtain the number of seeds in each seed hole; and

[0024] S303, the total number of seeds identified is used to calculate the single seed rate, reseeding rate and missed seeding rate, and the coordinates of the missed seeding holes are obtained as data information for later reseeding.

[0025] The beneficial effects of the present invention are:

[0026] The seedling tray conveying line of the present invention can match the conveyor belt speed with the seedling precision seeding production line, and can be raised and lowered to ensure the continuity of the entire operation process; the camera position and angle adjustment mechanism can change the shooting position and angle of the hole tray image acquisition camera, and combined with the light intensity adjustment mechanism, the captured picture can achieve a shooting effect that meets the detection requirements; the seedling precision seeding control system modularizes the functions, which can not only realize the adjustment and control of the device, but also integrate advanced target detection algorithms, exchange information with other control equipment (such as seedling precision seeding production line controllers), and ensure the collaborative operation of the device. In addition, the editable and backup functions of the database can facilitate the direct export of operation statistics, reduce manual labor intensity and avoid manual recording errors.

[0027] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments, but this does not limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 This is a schematic structural diagram of a detection device according to an embodiment of the present invention;

[0029] Figure 2 A cross-sectional view of a detection box according to an embodiment of the present invention;

[0030] Figure 3 An exploded view of a detection box according to an embodiment of the present invention;

[0031] Figure 4 A schematic structural diagram of a camera adjustment mechanism according to an embodiment of the present invention;

[0032] Figure 5 It is a working principle diagram of the present invention;

[0033] Figure 6 FIG. 4 is a flow chart of a target detection algorithm according to an embodiment of the present invention.

[0034] Among them, the reference numerals

[0035] 1 supporting leg

[0036] 2Seedling tray conveying mechanism

[0037] 21 Delivery Stent

[0038] 22 conveyor belt

[0039] 23 conveyor drive motor

[0040] 24 conveyor control cabinet

[0041] 25 limit rod

[0042] 3. Seedling tray position detection mechanism

[0043] 4 detection box

[0044] 41 box rack

[0045] 42 visor

[0046] 43 bar light sources

[0047] 44 cameras

[0048] 441 lens

[0049] 5. Detection and control mechanism

[0050] 51 touch screen

[0051] 6Camera adjustment mechanism

[0052] 61 Normal Angle Adjustment Servo

[0053] 62 horizontal angle adjustment servo

[0054] 63 front and rear adjustment drive motor

[0055] 64 front and rear adjustment sliders

[0056] 65 front and rear adjustment rails

[0057] 66 height adjustment slider

[0058] 67 Height adjustment fixture

[0059] 68 height adjustment slide rail

[0060] 69 Height adjustment drive motor DETAILED DESCRIPTION

[0061] The structural principle and working principle of the present invention are described in detail below with reference to the accompanying drawings:

[0062] See also Figure 1 , Figure 1The structure diagram of the detection device of one embodiment of the present invention. The detection device for precision seeding of seedlings of the present invention comprises: a support leg 1, which is preferably a liftable structure to adjust the height of the conveyor belt 22 to be consistent with the level of the conveyor belt 22 of the seedling sowing production line; a seedling tray conveying mechanism 2, which comprises a conveying bracket 21, a conveyor belt 22, a conveying drive motor 23 and a conveying control cabinet 24. The conveying bracket 21 is mounted on the support leg 1, and the conveyor belt 22, the conveying drive motor 23 and the conveying control cabinet 24 are respectively mounted on the conveying bracket 21. The conveying motor is connected to the conveyor belt 22 and the conveying control cabinet 24 respectively, and the speed can be adjusted by the conveying control cabinet 24, so as to facilitate the precision seeding of seedlings. The sowing production line is integrated, and different types of communication serial ports are provided at the end of the conveying control cabinet 24 to meet the communication requirements between different devices to better achieve integration; a detection box 4 is installed above the conveying bracket 21, and an image acquisition module is provided in the detection box 4 corresponding to the conveyor belt 22; and a detection control mechanism 5 is installed on the detection box 4 and connected to the conveying control cabinet 24, which is used for image acquisition of the hole tray after sowing and sowing situation analysis and recording during the seedling sowing process, and a deep learning model is used to post-process the acquired image to obtain sowing information, and the sowing information is stored in a local database.

[0063] The detection control mechanism 5 includes a speed adjustment unit, a light intensity adjustment unit, an image acquisition unit, a camera 44 position control unit, an image processing unit, a human-computer interaction unit, a data analysis unit, a data storage unit, a data transmission unit, and a serial port expansion unit. The serial port expansion unit is used for communication between different devices. The data storage unit stores default detection parameters. The data analysis unit uses a target detection algorithm to identify and detect crops and automatically matches image information obtained from seed sowing analysis. It may also include a touch screen 51, which can be used to set operating parameters. To facilitate operator operation and acquisition of detection information, the touch screen 51 is preferably mounted on the top of the detection box 4 and can be freely adjusted in angle. Data processing statistical results can be displayed on the touch screen 51, including the captured image and single seed rate, replay rate, and missed seed rate. The human-computer interaction unit connects to the touch screen 51 to perform visual operations and edit the data storage format, providing control over interactive operations and displaying detection statistical information. The image acquisition unit and image processing unit are responsible for running the algorithm model for hole tray identification and detection. The data storage unit, data transmission unit, and serial port expansion unit provide common interfaces such as a pinout, USB communication interface, and RJ45 interface, facilitating functional expansion and information exchange between different devices and hardware. The RJ45 interface can be used to communicate with the controller installed in the seedling and sowing production line, facilitating collaborative control and integration; the USB interface is used for data copying and manual updates; the HDMI interface is used for communication with the human-machine interface (touch screen 51); other communication protocols such as Wi-Fi are used for remote detection, communication, and data transmission, and CAN is used for later control function expansion, such as communication and control of hardware such as electrical proportional control valves to achieve adaptive pressure control.

[0064] The detection control mechanism 5 can have a built-in database to store default detection parameters, and the detection parameters set last time are preferably used when the operation starts. The detection control mechanism 5 and the conveying control cabinet 24 share a built-in database, and can share data with expansion equipment (PLC controller, etc.) through the communication serial port to achieve data-driven system integration. The user's copy operation of the data can be implemented through the USB interface of the detection control mechanism 5. The data to be saved and the format to be saved are selected through the touch screen 51, and Excel's .xls and .xlsx formats are supported. The seedling tray target detection model is copied to the built-in memory card through the USB interface and automatically loaded during operation.

[0065] In this embodiment, the seedling tray conveying mechanism 2 also includes a limit rod 25, which is installed on the conveying racks on both sides of the conveyor belt 22. The limit rod 25 is used to constrain the lateral position of the seedling tray so that it does not deviate or twist in the lateral position, and ensure that the seedling tray is in the same orientation during the forward conveying process so that the collected seedling tray images are consistent, which facilitates the smooth progress of the detection work; seedling trays of different specifications are constrained by adjusting the spacing of the limit rods 25. Among them, it can also include a seedling tray position detection mechanism 3, which is installed on the limit rod 25 and connected to the detection control mechanism 5, and forms an adjustable angle of 30° to 60° with the forward direction of the conveyor belt 22, and is used to detect whether there is a seedling tray in front and trigger the picture acquisition instruction, that is, the image acquisition time is determined by the seedling tray position detection mechanism 3. The seedling tray position detection mechanism 3 can be a detection sensor, preferably a high-precision laser photoelectric switch.

[0066] See also Figure 2 and Figure 3 , Figure 2 This is a cross-sectional view of a detection box 4 according to an embodiment of the present invention. Figure 3 The exploded view of the detection box 4 of one embodiment of the present invention. The detection box 4 of this embodiment includes a box frame 41 and a sunshade 42 wrapped on the box frame 41. The sunshade 42 can isolate the influence of external environmental light and reflect and enhance the internal light to ensure that the light covers the entire seedling tray and the imaging is evenly bright and dark. The image acquisition module includes a camera 44. In order to make the imaging clearer and meet the detection requirements of the target detection algorithm, according to the different types and characteristics of the sown seeds, the camera 44 that meets the technical requirements can be replaced, such as: thermal imaging camera 44, depth camera 44, color array camera 44, etc., to make the imaging information richer. The lens 441 of the camera 44 can be selected to replace lenses 441 with different focal lengths, such as 4mm lens 441, 6mm lens 441, 8mm lens 44. 1, 16mm lens 441, etc., to ensure that the seedling tray image is completely and clearly collected under different height conditions; the detection box 4 is also provided with a camera adjustment mechanism 6 and a strip light source 43, the camera adjustment mechanism 6 is installed at the upper end of the box frame 41 of the detection box 4, and the camera 44 is installed on the camera adjustment mechanism 6 to adjust the shooting position and angle; the shooting light intensity affects the exposure time of the camera 44 and the final imaging effect, and the strip light source 43 can adjust the light intensity in the detection box 4 through the light intensity adjustment unit to achieve the condition of clear image; the installation direction of the strip light source 43 is parallel to the forward direction of the conveyor belt 22, that is, parallel to the seedling tray conveying direction, and is tilted downward to illuminate the conveyor belt 22.

[0067] See also Figure 4 , Figure 4Figure 6 is a schematic diagram of the camera adjustment mechanism 6 according to one embodiment of the present invention. To accommodate the inspection requirements for seedling trays of varying sizes, in addition to the adjustable spacing of the limit rods 25, the camera adjustment mechanism 6 can also adjust the height and shooting angle of the camera 44, determining the appropriate position and angle of the image acquisition camera 44 to ensure that the image meets the inspection requirements. The camera adjustment mechanism 6 of this embodiment includes a normal angle adjustment servo 61, a horizontal angle adjustment servo 62, a front-to-back adjustment slide and a height adjustment slide. The front-to-back adjustment slide includes a front-to-back adjustment drive motor 63, a front-to-back adjustment slide rail 65 and a front-to-back adjustment slider 64. The front-to-back adjustment drive motor 63 is connected to the front-to-back adjustment slide rail 65, and the front-to-back adjustment slider 64 is installed on the front-to-back adjustment slide rail 65; the height adjustment slide includes a height adjustment drive motor 69, a height adjustment slide rail 68 and a height adjustment slider 66. The height adjustment drive motor 69 is connected to the height adjustment slide rail 68, and the height adjustment slider 66 is installed on the height adjustment slide rail 68; the front-to-back adjustment slide rail 65 is connected to the height adjustment slider 66; the height adjustment slide rail 68 is connected to the box frame 41 of the detection box 4 through the height adjustment fixing member 67; the camera 44 is connected to the front-to-back adjustment slider 64 through a two-degree-of-freedom pan-tilt platform composed of the normal angle adjustment servo 61 and the horizontal angle adjustment servo 62. Among them, the normal angle adjustment servo 61 adjusts the shooting angle of the camera 44 from 0 to 120° in the normal plane of the conveying plane of the conveyor belt 22; the horizontal angle adjustment servo 62 adjusts the shooting angle of the camera 44 from 0 to 360° in the parallel plane of the conveying plane of the conveyor belt 22.

[0068] The present invention is suitable for seeding quality inspection in general vegetable seedling precision seeding production lines. Installed and integrated between the seeding and soil covering stages, it detects and calculates the single seed rate, reseeding rate, missed seeding rate, and the coordinate positions of missed seeding holes in seedling trays, and stores the data in a local database. The seedling tray conveyor mechanism 2 adjusts the height of the conveyor belt 22 via a liftable support leg 1; the inspection box 4 changes the position of the camera 44 via a camera adjustment mechanism 6, and the image acquisition time is determined by the seedling tray position detection mechanism 3; the detection control mechanism 5 displays a human-computer interaction interface via a touch screen 51, allowing the configuration of relevant parameters such as detection tasks and data transmission.

[0069] See also Figure 5 , Figure 5 The working principle diagram of the present invention is shown in FIG. The seedling precision sowing detection method of the present invention uses the above-mentioned seedling precision sowing detection device to collect and analyze the hole tray sowing situation between the sowing and covering stages of the seedling sowing production line, and includes the following steps:

[0070] Step S100, initialization of detection parameters, loading the most recently used detection parameters retained in the local database or loading the default detection parameters (if it is the first time to use, the default detection parameters are loaded), and the detection parameters can be reset through the human-computer interaction interface; the detection parameters include: seed category: optional bulk vegetable crops: peppers, eggplants, tomatoes, cucumbers, etc.; plug tray specifications: optional standard plug trays with 72 holes, 128 holes, and 256 holes; collection interval: optional parameters 0 to 3 trays, indicating the interval between the plug trays for shooting and collection during operation, the default is 0, indicating continuous collection; light intensity: optional parameters 0 to 100%, adjusting the brightness of the strip light source 43 of the collection box; camera 44 posture parameters: front and rear position 100 to 600 mm, height position 0 to 450 mm, rotation angle (shooting angle adjustment 0 to 120° in the normal plane of the conveying plane, shooting angle adjustment 0 to 360° in the parallel plane of the conveying plane), keep the captured picture clear and centered;

[0071] Step S200, image acquisition: During the seedling raising and sowing process, the seedling tray position detection mechanism 3 triggers the camera 44 to capture images of the plug tray after sowing. For example, the timing of the camera 44 capturing images can be determined by whether the seedling tray position detection mechanism 3 (e.g., a laser infrared switch) is triggered.

[0072] Step S300: using a deep learning model to post-process the collected image to obtain seeding information; and

[0073] Step S400: storing the sowing information in a local database, wherein the sowing information includes a sowing single seed rate, a reseeding rate, a missed seed rate, and position coordinates of missed seed cells.

[0074] See also Figure 6 , Figure 6 This is a flow chart of a target detection algorithm according to an embodiment of the present invention. In step S300 of this embodiment, the deep learning model is a target detection algorithm model trained for the collected datasets of different seeds, and the processing after image collection further includes:

[0075] Step S301: After grayscale preprocessing, image morphological noise reduction, and Hough transform, the collected image is divided and cropped according to the wells (i.e., individual seed holes) to obtain the image and position coordinates (row and column positions) of the individual wells;

[0076] Step S302: calling the target detection algorithm model to perform identification and obtain the number of seeds in each seed hole; and

[0077] Step S303: The total number of seeds identified is used to calculate the single seed rate, reseeding rate and missed seeding rate, and the coordinates of the missed seeding holes are obtained as data information for subsequent reseeding.

[0078] The seedling precision seeding detection device of the present invention can be connected in series between the sowing and covering stages of a seedling production line. It can collect and analyze the seeding conditions in the seeding trays and post-process the captured images using a deep learning model. This allows for image capture of the seeding trays after sowing, as well as analysis and recording of the sowing conditions, during the seedling production process. It can obtain sowing information such as the single seed rate, reseeding rate, missed seeding rate, and the location coordinates of the missed seeding cells, and store the resulting data in a local database. This provides technical support for statistical analysis of the seedling production line's operations and subsequent reseeding.

[0079] Of course, the present invention may have many other embodiments. Without departing from the spirit and essence of the present invention, those skilled in the art may make various corresponding changes and modifications based on the present invention, but these corresponding changes and modifications should all fall within the scope of protection of the claims attached to the present invention.

Claims

1. A seedling precision sowing detection device, characterized in that: include: Support leg; The seedling tray conveying mechanism includes a conveying bracket, a conveyor belt, a conveying drive motor and a conveying control cabinet. The conveying bracket is installed on the support leg, and the conveyor belt, the conveying drive motor and the conveying control cabinet are respectively installed on the conveying bracket. The conveying motor is connected to the conveyor belt and the conveying control cabinet respectively. A detection box is installed above the conveying bracket, and an image acquisition module is provided in the detection box corresponding to the conveyor belt; as well as The detection control mechanism is installed on the detection box and connected to the conveying control cabinet. It is used to collect images of the hole tray after sowing and analyze and record the sowing conditions during the seedling sowing process. A deep learning model is used to post-process the collected images to obtain sowing information, and the sowing information is stored in a local database.

2. The seedling raising precision sowing detection device according to claim 1, characterized in that: The detection control mechanism includes a speed adjustment unit, a light intensity adjustment unit, an image acquisition unit, a camera posture control unit, an image processing unit, a human-computer interaction unit, a data analysis unit, a data storage unit, a data transmission unit and a serial port expansion unit. The serial port expansion unit is used for communication and interaction between different devices; the data storage unit stores detection default setting parameters, and the data analysis unit uses a target detection algorithm to realize crop recognition and detection, and automatically matches the image information obtained by sowing seed analysis.

3. The seedling raising precision sowing detection device according to claim 1, characterized in that: The supporting legs are of a liftable structure so as to adjust the height of the conveyor belt to be consistent with the level of the conveyor belt of the seedling and sowing production line.

4. The seedling raising precision sowing detection device according to claim 1, characterized in that: The seedling tray conveying mechanism also includes a limit rod, which is installed on the conveying frame on both sides of the conveyor belt. The limit rod is used to constrain the seedling tray from lateral position deviation and twisting, and to ensure that the seedling tray posture is consistent during the forward process so that the collected seedling tray images are consistent; seedling trays of different specifications are constrained by adjusting the spacing of the limit rods.

5. The seedling raising precision sowing detection device according to claim 4, characterized in that: It also includes a seedling tray position detection mechanism, which is installed on the limit rod and connected to the detection control mechanism, forming an adjustable angle of 30° to 60° with the forward direction of the conveyor belt, and is used to detect whether the seedling tray is transported in place and trigger a picture collection instruction.

6. The seedling raising precision sowing detection device according to claim 1, characterized in that: The detection box includes a box frame and a light shielding plate covered on the box frame; the image acquisition module includes a camera, and a camera adjustment mechanism and a strip light source are also provided in the detection box. The camera adjustment mechanism is installed at the upper end of the box frame of the detection box, and the camera is installed on the camera adjustment mechanism to adjust the shooting position and angle; the installation direction of the strip light source is parallel to the forward direction of the conveyor belt, and it is tilted downward toward the conveyor belt.

7. The seedling raising precision sowing detection device according to claim 6, characterized in that: The camera adjustment mechanism includes a normal angle adjustment servo, a horizontal angle adjustment servo, a front and rear adjustment slide and a height adjustment slide, the front and rear adjustment slide includes a front and rear adjustment drive motor, a front and rear adjustment slide rail and a front and rear adjustment slider, the front and rear adjustment drive motor is connected to the front and rear adjustment slide rail, and the front and rear adjustment slider is installed on the front and rear adjustment slide rail; the height adjustment slide includes a height adjustment drive motor, a height adjustment slide rail and a height adjustment slider, the height adjustment drive motor is connected to the height adjustment slide rail, and the height adjustment slider is installed on the height adjustment slide rail; the front and rear adjustment slide rail is connected to the height adjustment slider; the height adjustment slide rail is connected to the box frame of the detection box through the height adjustment fixing piece; the camera is connected to the front and rear adjustment slider through a two-degree-of-freedom pan-tilt composed of the normal angle adjustment servo and the horizontal angle adjustment servo.

8. The seedling raising precision sowing detection device according to claim 7, characterized in that: The normal angle adjustment servo adjusts the shooting angle of the camera to 0 to 120° in the normal plane of the conveyor belt conveying plane; the horizontal angle adjustment servo adjusts the shooting angle of the camera to 0 to 360° in the parallel plane of the conveyor belt conveying plane.

9. A method for detecting seedling precision sowing, characterized in that: The seedling precision sowing detection device according to any one of claims 1 to 8 is used to collect and analyze the hole tray sowing situation between the sowing and soil covering stages of the seedling sowing production line, comprising the following steps: S100, initializing detection parameters, loading the most recently used detection parameters stored in the local database or loading the default detection parameters, and resetting the detection parameters through the human-computer interaction interface; S200, image acquisition, during the seedling raising and sowing process, the seedling tray position detection mechanism triggers the camera to capture images of the seedling tray after sowing; S300, using a deep learning model to post-process the collected image to obtain seeding information; as well as S400: Storing the sowing information in a local database, the sowing information including a sowing single seed rate, a reseeding rate, a missed seeding rate, and position coordinates of missed seeding cells.

10. The seedling raising precision sowing detection method according to claim 9, characterized in that: In step S300, the deep learning model is a target detection algorithm model trained for the collected data sets of different seeds. The processing after image collection further includes: S301, after grayscale preprocessing, image morphological noise reduction, and Hough transform, the collected image is cropped according to the grid division to obtain the image and position coordinates of a single grid; S302, calling the target detection algorithm model to perform identification and obtain the number of seeds in each seed hole; and S303, the total number of seeds identified is used to calculate the single seed rate, reseeding rate and missed seeding rate, and the coordinates of the missed seeding holes are obtained as data information for later reseeding.

Images