An automated and rapid screening method and system for maize haploid kernels

Abstract

This invention belongs to the field of screening haploid corn kernels, specifically relating to an automated and rapid screening method and system for haploid corn kernels. The method includes: corn kernels entering the first track individually from a material bin; acquiring a top-view image of the corn kernels located on the first track as a first image; determining whether the germ surface of the corn kernel is facing upwards or downwards based on a germ surface detection and recognition model; if the germ surface is upwards, screening out haploid and diploid kernels based on the haploid detection and recognition model; if the germ surface is downwards, flipping the corn kernels 180° using a flipping mechanism to reach the beginning of a second track; acquiring a top-view image of the flipped corn kernels as a second image; screening out haploid and diploid kernels based on the haploid detection and recognition model; and outputting the haploid corn kernels. This invention can identify and screen haploid kernels, is simple to operate, automatically flips the germ surface of corn kernels, and automates the screening method, improving work efficiency.

Description

Technical Field

[0001] This invention belongs to the field of screening technology for maize haploid kernels, specifically relating to an automated and rapid screening method and system for maize haploid kernels. Background Technology

[0002] Currently, the methods used for automated detection and sorting of haploid corn kernels are genetic marker methods and oil content marker methods, corresponding to machine vision and nuclear magnetic resonance (NMR), respectively. Online NMR oil-containing seed sorting systems measure the oil content of each corn seed and distinguish between haploid and diploid kernels based on this content, achieving a 94% accuracy rate. The scanning time per seed is 4 seconds, but the system is large, expensive, and costly. Machine vision equipment offers higher accuracy and a better price-performance ratio. However, existing machine vision sorting systems are cumbersome to operate and inconvenient to maintain; they are not fully automated, requiring manual alignment of the kernels with the embryo surface facing upwards, resulting in low overall efficiency; and their self-learning ability is low, making them prone to errors. Summary of the Invention

[0003] The technical problem to be solved by this invention is the low level of automation in existing maize haploid kernel screening.

[0004] The technical solution adopted by this invention to solve its technical problem is: an automated and rapid screening method for haploid maize kernels, comprising:

[0005] S1, corn kernels enter the first track one by one from the material box;

[0006] S2, a first color camera installed above the first track, captures a top-view image of corn kernels located on the first track, as the first image;

[0007] S3. Based on the first image in S2, and according to the germ surface detection and recognition model, determine whether the germ surface of the corn kernel is facing up or down. If the germ surface of the corn kernel is facing up, then execute S4; if the germ surface of the corn kernel is facing down, then execute S5.

[0008] S4. Using the haploid detection and identification model, the first image of corn kernels with the germ surface facing up is used to screen out haploids and diploids.

[0009] S5, the corn kernels located at the end of the first track with the germ facing down are flipped 180° up and down by the flipping mechanism and then reach the beginning of the second track.

[0010] S6, a second color camera mounted above the second track, captures a top-down image of the corn kernels after they have been flipped on the second track, as the second image;

[0011] S7. Based on the second image in S6, haploids and diploids are selected according to the haploid detection and recognition model.

[0012] S8 outputs haploid corn kernels from S4 and S7.

[0013] Furthermore, the establishment of the embryo surface detection and recognition model is based on a first basic database composed of embryo surface upward and downward basic data extracted from the sample image, combined with a preset first basic model, which is a deep convolutional neural network model.

[0014] Furthermore, the establishment of the haploid detection and recognition model is based on a second basic database composed of basic data on the presence or absence of color in haploid and diploid corn germ surfaces extracted from sample images, combined with a preset second basic model, which is a deep convolutional neural network model.

[0015] Furthermore, S3 also includes filtering out irregularly shaped corn kernels based on the similarity between the first image in S2 and a preset standard image.

[0016] Furthermore, in S5, the structure of the flipping mechanism includes a first flipping wheel, a second flipping wheel, a flipping belt, a driven wheel assembly, a first cover belt, and a second cover belt; the flipping belt is configured to cooperate with the first flipping wheel and the second flipping wheel.

[0017] Furthermore, in S4, based on the first image, the presence or absence of color on the germ surface is identified using the haploid detection and recognition model, and it is determined whether the corn kernel corresponding to the first image is haploid or diploid.

[0018] Furthermore, S7 also includes filtering out irregularly shaped corn kernels based on the similarity between the second image in S6 and the preset standard image.

[0019] An automated rapid screening system for haploid maize kernels includes:

[0020] The modeling module is used to construct a first basic database based on the upward and downward orientation of the germ surface in the sample images, and to establish a germ surface detection and recognition model using a deep convolutional neural network model as the first basic model; and to construct a second basic database based on the presence or absence of color on the germ surface of haploid and diploid maize in the sample images, and to establish a haploid detection and recognition model using a deep convolutional neural network model as the second basic model.

[0021] The storage module is used to store the embryo surface detection and identification model and the haploid detection and identification model in the modeling module.

[0022] The feeding module is used to output corn kernels one by one and at equal intervals from the beginning of the first track to the first conveying module;

[0023] The first conveying module, which is the first track, is used to convey corn kernels to the position below the first collection module;

[0024] The first acquisition module is used to acquire images of the corn kernels from above within a predetermined area and send them to the first processing module;

[0025] The first processing module is used to determine whether the germ surface of the corn kernel is facing upward or downward according to the germ surface detection and recognition model in the storage module; and to filter out corn kernels with the germ surface facing upward as haploid corn kernels and diploid corn kernels according to the haploid detection and recognition model in the storage module.

[0026] The first screening module is used to screen corn kernels with the germ side facing up and diploid corn kernels with the germ side facing up according to the nozzles of the air pump connected by the solenoid valve. The germ side facing down is blown into different collection boxes. The corn kernels with the germ side facing down are output to the flipping module.

[0027] The flipping module is used to flip the corn kernels located at the end of the first track with the germ facing down by 180° up and down to the beginning of the second track.

[0028] The second conveying module, which is the second track, is used to convey the flipped corn kernels to the position below the second collection module;

[0029] The second acquisition module is used to acquire images of the top view of the flipped corn kernels within a predetermined area and send them to the second processing module.

[0030] The second processing module is used to filter out haploid and diploid cells from the top view image of the flipped corn kernels based on the haploid detection and recognition model in the storage module.

[0031] The second screening module is used to screen the haploid and diploid corn kernels classified by the second processing module according to the nozzles connected to the air pump via the solenoid valve. The haploid and diploid corn kernels will be blown into different collection boxes.

[0032] The output module is used to output haploid corn kernels from the first and second screening modules.

[0033] The beneficial effects of this invention are as follows: The automated rapid screening method and system for corn haploid kernels of this invention outputs single corn kernels onto a first track, and a first color camera located above the first track acquires a top-view first image of the corn kernels and identifies the orientation of the germ surface of the corn kernels. If a top-view image of the corn kernels facing upwards is acquired, the germ surface of the corn kernels is rotated 180° up and down by a flipping mechanism. A second color camera located above the second track acquires a top-view second image of the corn kernels. Based on the haploid detection model, haploid kernels and diploid kernels are identified and screened. The operation is relatively simple, realizes automatic flipping of the corn kernel germ surface, makes the screening method more automated, and improves work efficiency. Attached Figure Description

[0034] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0035] Figure 1 This is a flowchart of an automated rapid screening method for maize haploid kernels provided according to the first embodiment of the present invention;

[0036] Figure 2 This is a schematic diagram of the structure of an automated rapid screening system for maize haploid kernels provided according to the second embodiment of the present invention;

[0037] Figure 3 This is a schematic diagram of the flipping mechanism in an automated rapid screening system for maize haploid kernels provided according to the second embodiment of the present invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0039] The first embodiment of the present invention relates to an automated and rapid screening method for corn haploid kernels. In this embodiment, corn kernels are first output individually from a material box onto a predetermined track, which is the first track. A first color camera located above the first track acquires a top-view first image of the corn kernels and identifies the orientation of the germ surface. If a top-view image of the corn kernels with the germ surface facing upwards is acquired, haploid and diploid kernels are identified and screened according to the haploid detection model. If a top-view image of the corn kernels with the germ surface facing downwards is acquired, the germ surface of the corn kernels is rotated 180° up and down by a flipping mechanism, and a second color camera located above the second track acquires a top-view second image of the corn kernels. Haploid and diploid kernels are identified and screened according to the haploid detection model. The operation is simple, realizes automatic flipping of the corn kernel germ surface, and makes the screening method more automated, thus improving work efficiency.

[0040] The following details the implementation of the automated rapid screening method for maize haploid kernels in this embodiment. The following details are provided for ease of understanding and are not essential for implementing this solution. The specific process of this embodiment is as follows: Figure 1 As shown:

[0041] Step S1: Corn kernels are individually fed from the material box into the beginning of the first track;

[0042] Specifically, the corn kernels located in the material box are vibrated at the bottom outlet of the material box, and enter the starting end of the first track in a row of equidistant individual kernels at the outlet.

[0043] Step S2: The first color camera installed above the first track captures a top-view image of the corn kernels located on the first track;

[0044] Specifically, a first color camera with a constant light source is fixed directly above the first track to capture an image of the corn kernels from above, which serves as the first image.

[0045] Step S3: Based on the first image in step S2, determine whether the germ surface of the corn kernel is facing up or down according to the germ surface detection and recognition model. If the germ surface of the corn kernel is facing up, proceed to step S4; if the germ surface of the corn kernel is facing down, proceed to step S5.

[0046] Specifically, the first image of the collected corn kernels may be corn kernel A with the germ facing upwards or corn kernel B with the germ facing downwards. The texture and shape of corn kernel A with the germ facing upwards and corn kernel B with the germ facing downwards differ. Based on a first basic database constructed from the basic data of germ-facing orientation extracted from the sample images, and combined with this first basic database, 1200 extracted actual images are first preprocessed as sample images. First, orientation normalization is performed on the 1200 sample images. Then, the orientation-normalized 1200 sample images are used to train a germ-facing detection and recognition model. Afterwards, the first image is orientation-normalized and classified using the germ-facing detection and recognition model, making the corn kernel identification results more accurate. Based on a pre-defined deep convolutional network model, the trained germ-facing detection and recognition model mainly classifies corn kernels into three categories: corn kernels with the germ facing upwards, corn kernels with the germ facing downwards, and irregularly shaped corn kernels. The normalized first image is input into the germ surface detection and recognition model, which outputs the classification result of the first image, filtering out corn kernels A with the germ surface facing upward, corn kernels B with the germ surface facing downward, and irregularly shaped corn kernels.

[0047] Step S4: Using the haploid detection and identification model, the first image of corn kernels with the germ surface facing up is used to screen out haploids or diploids.

[0048] Specifically, the first image of corn kernels A with the germ surface facing upwards, as classified in step S3, is further processed. The color of the germ surface can distinguish whether the corn kernel is haploid or diploid. A second basic database is constructed based on the fundamental data extracted from the sample images regarding the presence or absence of color on the germ surfaces of haploid and diploid corn kernels. First, 1200 extracted actual images are used as sample images for preprocessing. First, the 1200 sample images undergo orientation normalization. Then, using the orientation-normalized 1200 sample images, a haploid detection and recognition model is trained based on a pre-defined deep convolutional network model. Next, the first image of corn kernels A with the germ surface facing upwards is further filtered using the haploid detection and recognition model to identify haploid and diploid corn kernels. The haploid and diploid corn kernels are then blown into different collection boxes via nozzles on an air pump connected to solenoid valves.

[0049] Step S5: The corn kernels located at the end of the first track with the germ facing down are flipped 180° up and down by the flipping mechanism and then reach the beginning of the second track.

[0050] Specifically, corn kernels B, located on the first track with the germ side down, are moved to the end of the first track and then slide into the corn kernel trough of the automatic flipping device. After passing through the rollers, due to gravity, corn kernels B are flipped 180° up and down, and then move to the beginning of the second track with the germ side up.

[0051] Step S6: The second color camera installed above the second track acquires a top-view image of the corn kernels after they have been flipped on the second track, as the second image;

[0052] Specifically, a second color camera with a constant light source is fixed directly above the second track to capture an image of the corn kernels from above, which serves as the second image;

[0053] Step S7: Based on the second image in S6, haploids or diploids are selected according to the haploid detection and recognition model;

[0054] Specifically, the top-view image of the flipped corn kernel B is used as the second image for further processing. The color of the germ surface can distinguish whether the corn kernel is haploid or diploid. A second basic database is constructed based on the color of the germ surface of haploid and diploid corn kernels extracted from the sample images. First, 1200 extracted actual images are used as sample images for preprocessing. First, the 1200 sample images undergo orientation normalization. Then, using the orientation-normalized 1200 sample images, a haploid detection and recognition model is trained based on a pre-defined deep convolutional network model. Next, the second image undergoes orientation normalization, and the haploid detection and recognition model is used to filter out haploid, diploid, and irregularly shaped corn kernels. The haploid, diploid, and irregularly shaped corn kernels are then blown into different collection boxes through nozzles of an air pump connected to solenoid valves.

[0055] Step S8: Output haploid corn kernels from S4 and S7.

[0056] The first embodiment of the present invention outputs a single corn kernel onto a predetermined track. A first color camera located above the first track acquires a top-down first image of the corn kernel and identifies the orientation of the germ surface. When a top-down image of the corn kernel facing upwards is acquired, haploid kernels are identified and screened based on a haploid detection model. When a top-down image of the corn kernel facing downwards is acquired, the germ surface of the corn kernel is rotated 180° up and down by a flipping mechanism. The rotated corn kernel passes through a second color camera located above the second track, acquiring a second top-down second image of the corn kernel. Haploid kernels are then identified and screened based on the haploid detection model. The operation is simple, achieving automatic rotation of the corn kernel germ surface, making the screening method more automated and improving work efficiency.

[0057] The steps of the various methods described above are only for clarity. In practice, they can be combined into one step or some steps can be split into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this patent. Adding insignificant modifications or introducing insignificant designs to the process, but without changing the core design of the algorithm and process, are also within the scope of protection of this patent.

[0058] like Figure 2 As shown, the second embodiment of the present invention relates to an automated rapid screening system for maize haploid kernels, comprising: a modeling module 201, a storage module 202, a feeding module 203, a first conveying module 204, a first processing module 205, a first processing module 206, a first screening module 207, a flipping module 208, a second conveying module 209, a second acquisition module 210, a second processing module 211, a second screening module 212, and an output module 213.

[0059] Specifically, the modeling module 201 is used to construct a first basic database based on whether the germ surface in the sample image is facing upwards or downwards, and to establish a germ surface detection and recognition model using a deep convolutional neural network model as the first basic model; and to construct a second basic database based on whether the germ surface of haploid and diploid corn in the sample image has color, and to establish a haploid detection and recognition model using a deep convolutional neural network model as the second basic model; the storage module 202 is used to store the germ surface detection and recognition model and the haploid detection and recognition model in the modeling module 201; and the feeding module 203 is used to feed corn kernels one kernel at a time from the beginning of the first track. The output is sent to the first conveying module 204; the first conveying module 204 is a first track used to convey corn kernels to a position below the first acquisition module 205; the first acquisition module 205 is used to acquire images of the top view of corn kernels within a predetermined area and send them to the first processing module 205; the first processing module 206 is used to determine whether the germ surface of the corn kernel is facing upward or downward according to the germ surface detection and recognition model in the storage module; according to the haploid detection and recognition model in the storage module, it filters out corn kernels with the germ surface facing upward as haploid corn kernels and diploid corn kernels; the first screening module 207 is used to... The corn kernels, sorted by the nozzle of the air pump connected to the solenoid valve, are separated into haploid and diploid corn kernels with the germ surface facing upwards and blown into different collection boxes. The corn kernels with the germ surface facing downwards are output to the flipping module 208. The flipping module 208 flips the corn kernels located at the end of the first track with the germ surface facing downwards by 180° to reach the beginning of the second track and outputs them to the second conveying module 209. The second conveying module 209, which is the second track, conveys the flipped corn kernels B to a position below the second collection module 210. The second collection module 210 is used for collecting... The system collects images of the top view of the flipped corn kernels within a predetermined area and sends them to the second processing module 211. The second processing module 211 is used to filter out haploid and diploid corn kernels from the top view images of the flipped corn kernels according to the haploid detection and recognition model in the storage module 202. The second screening module 212 is used to filter the haploid and diploid corn kernels classified by the second processing module 211 by using the nozzles of the air pump connected to the solenoid valve, and blows them into different collection boxes. The output module 213 is used to output the haploid corn kernels from the first screening module 207 and the second screening module 212.

[0060] Specifically, the flipping module 208 flips the corn kernels B with the germ side facing down through the flipping mechanism 1. The flipping mechanism 1 includes a feed inlet 11, a first flipping wheel 12, a second flipping wheel 13, a flipping belt 14, a driven wheel assembly 15, a first cover belt 16, a second cover belt 17, and a discharge outlet 18. The driven wheel assembly 15 includes a first driven wheel 151, a second driven wheel 152, a third driven wheel 153, a fourth driven wheel 154, a fifth driven wheel 155, and a sixth driven wheel 156. The flipping belt 14 has vertical through holes suitable for passing corn kernels. The flipping belt 14 cooperates with the first flipping wheel 12 and the second flipping wheel 13. The first driven wheel 11 and the second driven wheel 12 are fixed between the first flipping wheel 12 and the second flipping wheel 13. The first cover belt 16 passes through the first flipping wheel 12 and the first driven wheel 15 in sequence. 1. The second driven wheel 152 and the second rotating wheel 13 are then fixed around the first rotating wheel 12. The rotating belt 14 covers the inner side of the first rotating wheel 12 and the second rotating wheel 13. The third driven wheel 153, the fourth driven wheel 154, the fifth driven wheel 155, and the sixth driven wheel 156 are fixed around the outer side of the second rotating wheel 13. The second covering belt 17 is fixed around the third driven wheel 153, the second rotating wheel 13, the sixth driven wheel 156, the fifth driven wheel 155, and the fourth driven wheel 154 in sequence. The rotating belt 14 covers the outer side of the second rotating wheel 13. The feed inlet 18 is located directly above the rotating belt 14, and the discharge outlet 18 is located directly below the rotating belt 14. In use, the feed inlet 18 outputs single corn kernels into the through hole of the turning belt 14. The first covering belt 16 covers the underside of the turning belt 14, so that the corn kernels will not fall off the turning belt. The corn kernels enter the second turning wheel 13 along with the turning belt 14. The corn kernels in the through hole of the turning belt 14 on the second turning wheel 13 are covered by the first covering belt 16 and the second covering belt 17. Due to gravity, the corn kernels are turned 180 degrees and then enter the turning belt 14 located below. The corn kernels move to the point where they fall off the second covering belt 17 and enter the discharge port.

[0061] It is not difficult to see that this embodiment is a system implementation corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and will not be repeated here to reduce repetition. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.

[0062] It is worth mentioning that all modules involved in this embodiment are logical modules. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0063] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the scope of the present invention. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. An automated and rapid screening method for haploid maize kernels, characterized in that, include: S1, corn kernels enter the first track one by one from the material box; S2, a first color camera installed above the first track, captures a top-view image of corn kernels located on the first track, as the first image; S3. Based on the first image in S2, determine whether the germ surface of the corn kernel is facing up or down according to the germ surface detection and recognition model. If the germ surface of the corn kernel is facing up, then execute S4. If the corn germ is facing down, then execute S5; S4. Using the haploid detection and identification model, the first image of corn kernels with the germ surface facing up is used to screen out haploids and diploids. S5, the corn kernels located at the end of the first track with the germ facing down are flipped 180° up and down by the flipping mechanism and then reach the beginning of the second track. S6, a second color camera mounted above the second track, captures a top-down image of the corn kernels after they have been flipped on the second track, as the second image; S7. Based on the second image in S6, haploids and diploids are selected according to the haploid detection and recognition model. S8 outputs haploid corn kernels from S4 and haploid corn kernels from S7; The flipping mechanism comprises a first flipping wheel, a second flipping wheel, a flipping belt, a driven wheel assembly, a first covering belt, and a second covering belt. The flipping belt is configured to cooperate with the first and second flipping wheels. The driven wheel assembly includes a first driven wheel, a second driven wheel, a third driven wheel, a fourth driven wheel, a fifth driven wheel, and a sixth driven wheel. The flipping belt has vertical through holes suitable for passing corn kernels. The first and second driven wheels are fixed between the first and second flipping wheels. The first covering belt is fixed around the first flipping wheel, the first driven wheel, the second driven wheel, the second flipping wheel, and then around the first flipping wheel. The flipping belt covers the inner side of the first and second flipping wheels. The third, fourth, fifth, and sixth driven wheels are fixed around the outer side of the second flipping wheel. The second covering belt is fixed around the third driven wheel, the second flipping wheel, the sixth driven wheel, the fifth driven wheel, the fourth driven wheel, and then around the third driven wheel. The flipping belt covers the outer side of the second flipping wheel.

2. The automated rapid screening method for maize haploid kernels according to claim 1, characterized in that, The establishment of the embryo surface detection and recognition model is based on a first basic database composed of embryo surface upward and downward basic data extracted from the sample image, combined with a preset first basic model, which is a deep convolutional neural network model.

3. The automated rapid screening method for maize haploid kernels according to claim 1, characterized in that, The haploid detection and recognition model is established based on a second basic database composed of basic data on the presence or absence of color on the germ surface of haploid and diploid corn extracted from sample images, combined with a preset second basic model, which is a deep convolutional neural network model.

4. The automated rapid screening method for maize haploid kernels according to claim 1, characterized in that, S3 also includes filtering out irregularly shaped corn kernels based on the similarity between the first image in S2 and a preset standard image.

5. The automated rapid screening method for haploid maize kernels according to claim 1, characterized in that, In S4, based on the first image, the presence or absence of color on the germ surface is identified using the haploid detection and recognition model, and it is determined whether the corn kernel corresponding to the first image is haploid or diploid.

6. The automated rapid screening method for maize haploid kernels according to claim 1, characterized in that, S7 also includes filtering out irregularly shaped corn kernels based on the similarity between the second image in S6 and the preset standard image.

7. An automated rapid screening system for haploid maize kernels, characterized in that, include: The modeling module is used to construct a first basic database based on the upward and downward orientation of the germ surface in the sample images, and to establish a germ surface detection and recognition model using a deep convolutional neural network model as the first basic model; and to construct a second basic database based on the presence or absence of color on the germ surface of haploid and diploid maize in the sample images, and to establish a haploid detection and recognition model using a deep convolutional neural network model as the second basic model. The storage module is used to store the embryo surface detection and recognition model and the haploid detection and recognition model in the modeling module; The feeding module is used to output corn kernels one by one and at equal intervals from the beginning of the first track to the first conveying module; The first conveying module, which is the first track, is used to convey corn kernels to the position below the first collection module; The first acquisition module is used to acquire images of the corn kernels from above within a predetermined area and send them to the first processing module; The first processing module is used to determine whether the germ surface of the corn kernel is facing upward or downward according to the germ surface detection and recognition model in the storage module; and to filter out corn kernels with the germ surface facing upward as haploid corn kernels and diploid corn kernels according to the haploid detection and recognition model in the storage module. The first screening module is used to screen the haploid corn kernels with the germ surface facing up and the diploid corn kernels with the germ surface facing up according to the nozzles connected to the air pump via the solenoid valve, and blow them into different collection boxes. The corn kernels with the germ facing down are output to the flipping module; A flipping module is used to flip corn kernels located at the end of the first track with the germ side facing down by 180° up and down to the beginning of the second track. The flipping module uses a flipping mechanism to flip the corn kernels with the germ side facing down. The flipping mechanism includes a first flipping wheel, a second flipping wheel, a flipping belt, a driven wheel assembly, a first covering belt, and a second covering belt. The flipping belt is configured to cooperate with the first and second flipping wheels. The driven wheel assembly includes a first driven wheel, a second driven wheel, a third driven wheel, a fourth driven wheel, a fifth driven wheel, and a sixth driven wheel. The flipping belt has vertical through holes suitable for the passage of corn kernels. The first driven wheel... The first cover belt is fixed between the first rotating wheel and the second rotating wheel, and the second driven wheel is fixed by passing through the first rotating wheel, the first driven wheel, the second driven wheel, the second rotating wheel and then through the first rotating wheel in sequence. The rotating belt covers the inner side above the first rotating wheel and the second rotating wheel. The third driven wheel, the fourth driven wheel, the fifth driven wheel and the sixth driven wheel are fixed around the outer side of the second rotating wheel, and the second cover belt is fixed by passing through the third driven wheel, the second rotating wheel, the sixth driven wheel, the fifth driven wheel and the fourth driven wheel in sequence. The rotating belt covers the outer side of the second rotating wheel. The second conveying module, which is the second track, is used to convey the flipped corn kernels to the position below the second collection module; The second acquisition module is used to acquire images of the top view of the flipped corn kernels within a predetermined area and send them to the second processing module. The second processing module is used to filter out haploid and diploid cells from the top view image of the flipped corn kernels based on the haploid detection and recognition model in the storage module. The second screening module is used to screen the haploid and diploid corn kernels classified by the second processing module according to the nozzles connected to the air pump via the solenoid valve. The haploid and diploid corn kernels will be blown into different collection boxes. The output module is used to output haploid corn kernels from the first and second screening modules.

Images