Vegetable high-speed sorting method and system

Abstract

The invention provides a high-speed sorting method and system which are applied to the technical field of vegetable sorting, front and back images of to-be-sorted vegetables are obtained through shooting modules arranged above and below a food-grade transparent conveying belt, panoramic image data are formed, an artificial intelligence deep learning model based on a convolutional neural network is used for image analysis, and a high-speed sorting result is obtained. According to the vegetable sorting system, the defects in the vegetables are automatically recognized and classified, once the defects are found, the positions of the defects are accurately calculated, and removing operation is executed, so that efficient and accurate vegetable sorting is achieved, the sorting efficiency and accuracy are improved, the labor cost is reduced, the vegetable quality is improved, food safety is ensured, and the market competitiveness of enterprises is enhanced.

Description

Technical Field

[0001] The present application relates to the technical field of vegetable sorting, and in particular to a high-speed vegetable sorting method and system. Background Art

[0002] In the sorting equipment for vegetable sorting, traditional food processing sorting mainly relies on manual selection and industrial machines, among which manual sorting faces problems such as high cost, low efficiency, unstable quality and high labor intensity.

[0003] Industrial sorting machines mainly use color sorters. Mainstream color sorters rely on traditional color recognition algorithms and have limited overall recognition capabilities, especially in identifying foreign objects that are not related to color. In addition, color sorters can usually only shoot from above, and the shooting angle is limited to a maximum of 180 degrees, so they can only identify half of the material area and cannot capture the full picture of the material, resulting in incomplete detection of foreign objects and defects.

[0004] Based on this, a new high-speed sorting method, system and equipment are needed. Summary of the invention

[0005] In view of this, the embodiments of this specification provide a new high-speed sorting method, system and equipment, which are applied to the field of vegetable sorting technology, focusing on customer concerns such as performance optimization, food safety and product quality. Through 360-degree all-round scanning, foreign matter can be quickly and accurately detected and removed, and small defects of each product can be accurately identified in a large amount of feed. It not only improves the sorting efficiency and accuracy, but also effectively reduces labor costs, and improves product quality, ensures food safety, and enhances the company's market competitiveness.

[0006] The embodiments of this specification provide the following technical solutions:

[0007] The embodiment of this specification provides a high-speed vegetable sorting method, comprising:

[0008] The upper shooting module disposed above the food-grade transparent conveyor belt is used to obtain the front image of the vegetables to be sorted, and the lower shooting module disposed below the food-grade transparent conveyor belt is used to obtain the back image of the vegetables to be sorted;

[0009] Combining the front image and the back image into panoramic image data of the vegetables to be sorted;

[0010] Using an artificial intelligence deep learning model to identify the synthesized panoramic image data, and determine whether the vegetables to be sorted have defects;

[0011] If a defect exists, determining the type of the defect and calculating the defect position, wherein the defect position includes the position of the defect in the panoramic image data;

[0012] When vegetables with defects move to the execution area of ​​the high-speed spraying system, they are removed according to the type and location of the defects.

[0013] The embodiment of this specification also provides a high-speed vegetable sorting system, including: a vegetable sorting panoramic shooting system, a defect recognition system and a high-speed blowing system;

[0014] The vegetable sorting panoramic shooting system is used to obtain the front image and the back image of the vegetables to be sorted, and synthesize the front image and the back image into the panoramic image data of the vegetables to be sorted;

[0015] The front image of the vegetables to be sorted is obtained by an upper shooting module arranged above the food-grade transparent conveyor belt; the back image of the vegetables to be sorted is obtained by a lower shooting module arranged below the food-grade transparent conveyor belt;

[0016] The defect recognition system uses an artificial intelligence deep learning model to analyze the panoramic image data to determine whether the vegetables to be sorted have defects;

[0017] If a defect exists, determining the type of the defect and calculating the defect position, wherein the defect position includes the position of the defect in the panoramic image data;

[0018] The high-speed spraying system performs a rejection operation according to the type and position of the defect provided by the defect recognition system when the defective vegetable moves to the execution area of ​​the high-speed spraying system.

[0019] Compared with the prior art, the at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects:

[0020] The front and back images of the vegetables to be sorted are obtained by installing shooting modules above and below the food-grade transparent conveyor belt to form panoramic image data. The image analysis is then performed using an artificial intelligence deep learning model based on a convolutional neural network to automatically identify and classify defects in the vegetables. Once a defect is found, its location is accurately calculated and removed, thereby achieving efficient and accurate vegetable sorting. This not only improves sorting efficiency and accuracy and reduces labor costs, but also improves vegetable quality, ensures food safety, and enhances the company's market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0022] Figure 1 It is a schematic diagram of the dimensions of the high-speed vegetable sorting system in this application;

[0023] Figure 2 It is a structural schematic diagram of the vegetable sorting panoramic photography system of the present application;

[0024] Figure 3 is a structural schematic diagram of the shooting module in this application;

[0025] Figure 4 It is a partial enlarged view of the AA surface of the structural schematic diagram of the shooting module in this application;

[0026] Figure 5 It is a structural schematic diagram of the high-speed conveying equipment in this application;

[0027] Figure 6 It is a high-definition recognition image based on the artificial intelligence deep learning model in this application;

[0028] Figure 7 It is a schematic diagram of the network structure based on a convolutional neural network of the artificial intelligence deep learning model in this application;

[0029] Figure 8 It is a schematic diagram of the blowing compensation and blowing width calculation process in the high-speed blowing system of the present application;

[0030] Fig. 9 is the list of coordinate transformations in this application;

[0031] Fig.10 It is the high-speed blowing system of the present application;

[0032] In the figure: 1. fixing plate; 26. camera and lens; 28. light source angle adjustment plate; 29. ​​light source connecting plate; 32. light source mounting plate; 34. connecting plate; 38. left and right mounting plates; 43. handle; 44. protective cover; 46. high-brightness light source; 37. reduction motor; 36. motor anti-rotation structure; 51. left bearing seat; 48. left bearing; 62. seal; 23. main roller; 91. food-grade transparent conveyor belt; 58. right bearing seat; 55. right bearing; 11. structural carrier of high-speed blowing; 4. execution carrier of high-speed blowing; 21. protective box of high-speed blowing. DETAILED DESCRIPTION

[0033] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0034] The following describes the implementation methods of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the following embodiments and the features in the embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0035] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present application, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspect described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0036] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0037] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the examples can be practiced without these specific details.

[0038] At present, the main methods of food processing are manual and industrial sorting. Manual sorting has many problems, such as low efficiency, unstable quality, high labor intensity, and high cost for enterprises.

[0039] Industrial sorting mainly uses color sorters, which mainly rely on traditional color recognition algorithms. They can identify vegetables with abnormal colors, but perform poorly in identifying foreign objects, especially those with similar colors to vegetables, such as hair, toothpicks, plastic tape, sticky notes, grass, leaves, poplar fluff, big caterpillars, aphids, snails, ladybugs, earthworms, mosquitoes, flies, small flying insects, etc. mixed in the vegetables, making it difficult to effectively identify foreign objects and defects of the same color, and the overall recognition ability is limited.

[0040] In addition, mainstream color sorters can usually only shoot from top to bottom, with a limited shooting range of up to 180 degrees, which cannot capture the full picture of the material, making the detection of foreign matter and defects not thorough enough. At the same time, the conveyor belts of these color sorters run slowly, generally at 1-2 meters per second, which limits the overall sorting capacity.

[0041] In view of this, during the analysis and improvement exploration of this method, it was found that the current mainstream color sorters cannot fully identify foreign objects with similar colors to vegetables and foreign objects that are not related to color when sorting vegetables. In addition, the single-sided shooting method of the color sorter limits the recognition area, thereby increasing the risk of missing foreign objects and defects, and the slow conveyor belt further limits production capacity.

[0042] Based on this, the embodiment of this specification proposes a high-speed vegetable sorting method: Figure 2 As shown in the figure, the overall idea is: the front and back images of the vegetables to be sorted are obtained by setting up shooting modules above and below the food-grade transparent conveyor belt to form panoramic image data, and then the artificial intelligence deep learning model based on the ATSS algorithm is used for image analysis to automatically identify and classify defects in vegetables. Once a defect is detected, the type of defect is further determined and the location of the defect is accurately calculated, and finally it is removed, thereby achieving efficient and accurate vegetable sorting, which not only improves sorting efficiency and accuracy, reduces labor costs, but also improves vegetable quality and meets the needs of industrialized production.

[0043] The technical solutions provided by various embodiments of the present application are described below in conjunction with the accompanying drawings.

[0044] like Figure 2 As shown, the embodiment of this specification provides a vegetable high-speed sorting method, comprising:

[0045] The upper shooting module disposed above the food-grade transparent conveyor belt is used to obtain the front image of the vegetables to be sorted, and the lower shooting module disposed below the food-grade transparent conveyor belt is used to obtain the back image of the vegetables to be sorted;

[0046] Combining the front image and the back image into panoramic image data of the vegetables to be sorted;

[0047] Using an artificial intelligence deep learning model to identify the synthesized panoramic image data, and determine whether the vegetables to be sorted have defects;

[0048] If a defect exists, determining the type of the defect and calculating the defect position, wherein the defect position includes the position of the defect in the panoramic image data;

[0049] When vegetables with defects move to the execution area of ​​the high-speed spraying system, they are removed according to the type and location of the defects.

[0050] During implementation, the food-grade transparent conveyor belt smoothly transfers the materials (vegetables to be sorted) to the designated shooting area. The upper shooting module located above the food-grade transparent conveyor belt is responsible for shooting the vegetables to be sorted from above and capturing the front image of the vegetables to be sorted.

[0051] The lower shooting module located below the food-grade transparent conveyor belt works in conjunction with the upper shooting module to shoot the vegetables to be sorted from below and capture images of the back of the vegetables.

[0052] It should be emphasized that the food-grade transparent conveyor belt is a high-speed conveyor belt. Before the material enters the food-grade transparent conveyor belt, it will undergo pre-acceleration processing to match the speed of the conveyor belt, and a sufficient stable distance is set between the material entry point and the shooting point to ensure that the material and the food-grade transparent conveyor belt are relatively still when shooting, thereby capturing high-quality vegetable images, ensuring the clarity and accuracy of the image shooting, and providing accurate visual data for subsequent defect detection and foreign matter removal.

[0053] In addition, the food-grade transparent conveyor belt in the transparent conveying device can not only transport the materials smoothly to the designated shooting area, but also provide the required light transmittance and perspective effects for the lower shooting module below, so that the lower shooting module can clearly obtain the back image of the vegetables to be sorted, ensuring the shooting quality.

[0054] Then, the acquired front and back images of the vegetables to be sorted are synthesized into panoramic image data of the vegetables to be sorted, providing a comprehensive view of the vegetables for the artificial intelligence deep learning model.

[0055] Next, the trained artificial intelligence deep learning model is used to identify the synthesized panoramic image data to determine whether there are defects in the panoramic image;

[0056] For example, the defects of the vegetables themselves: yellow leaves, rotten leaves, burnt edges, and picking out rotten and deformed parts.

[0057] Another example is foreign objects and defects mixed in vegetables: hair, toothpicks, plastic tape, sticky notes, grass, leaves, poplar fluff, big caterpillars, aphids, snails, ladybugs, earthworms, mosquitoes, flies, small flying insects, etc.

[0058] If a defect is detected, it is classified to determine which category it belongs to, and image processing technology is used to accurately calculate the location of the defect on the vegetable. When the defective vegetable moves to the execution area of ​​the high-speed blowing system, the removal operation is performed based on the defect category and defect location information.

[0059] In one embodiment, the upper shooting module and the lower shooting module are staggeredly arranged above and below the food-grade transparent conveyor belt.

[0060] like Figure 2 As shown, the distance of the left and right misalignment is L. The distance between the two shooting modules is adjusted according to actual needs to ensure that every side of the vegetable can be photographed, covering the front and back of the vegetable, thereby achieving panoramic recognition without blind spots.

[0061] The staggered design of the upper and lower modules needs to take into account the field of view of the camera 26 and the moving speed of the vegetables on the conveyor belt to ensure the continuity and integrity of the image.

[0062] In one embodiment, the misalignment distance is not less than 1 / 2 of the field of view of the shooting module.

[0063] Specifically, the offset distance L is designed to be half of the field of view of the camera 26. When the vegetables pass through the conveyor belt, the upper module and the lower module can capture the front and back of the vegetables respectively, that is, covering the entire surface of the vegetables, without overlapping or omissions in the image, thereby improving the accuracy and efficiency of vegetable sorting.

[0064] The offset distance can be adjusted according to actual needs to adapt to different shooting scenes and materials.

[0065] In one embodiment, if Figure 3 and Figure 4 As shown, the upper shooting module and / or the lower shooting module includes a fill light unit, and the fill light unit includes a light source and an angle adjustment plate;

[0066] The light source is a high brightness light source 46, which is used to irradiate light to the vegetables to be sorted;

[0067] The angle adjustment plate is used to adjust the irradiation angle of the light source so that the light irradiated according to the irradiation angle covers the vegetables to be sorted.

[0068] Specifically, the light source is installed on the light source mounting plate 32 through the light source connecting plate 29 to ensure the stability of the light source. The high-brightness light source is responsible for providing sufficient brightness so that it can evenly and clearly illuminate the vegetables to be sorted under various lighting conditions, thereby enhancing the clarity and detail of the image, especially in the identification of vegetable surface features.

[0069] During implementation, in order to adapt to vegetables of various shapes and sizes to be sorted, the irradiation angle of the light source is finely adjusted through the light source angle adjustment plate 28, so as to control the direction and coverage of the light, reduce the shadow and reflection problems caused by the surface of the vegetables, and thus improve the visibility of the vegetable features, ensure the quality of image acquisition, and provide accurate data support for subsequent image processing and vegetable sorting.

[0070] In one embodiment, if Figure 5 As shown, the transparent conveying device further includes: a motor anti-rotation structure;

[0071] The motor anti-rotation structure cooperates with the reduction motor to accurately control the arrival of the vegetables to be sorted at a preset area, thereby increasing the running speed of the conveyor line and thus improving the sorting efficiency and overall production capacity.

[0072] like Figure 5 As shown, the motor anti-rotation structure 36 fixes the reduction motor 37 and drives the motor to center, thereby preventing the motor from rotating or shifting during operation, and accurately controlling the vegetables to be sorted to reach the preset shooting area and the high-speed spray system execution area.

[0073] In one embodiment, the food-grade transparent conveyor belt 91 is supported by the left bearing seat 51, the right bearing seat 58 and the left bearing 48, the right bearing 55, thereby ensuring that the food-grade transparent conveyor belt 91 runs smoothly under the drive of the main roller 23.

[0074] Preferably, sealing grooves are provided on the left bearing seat 51 and the right bearing seat 58 for installing a seal 62 to achieve dustproof and waterproof functions and protect the left bearing 48 and the right bearing 55 from the influence of the external environment.

[0075] In one embodiment, the artificial intelligence deep learning model is used to identify the synthesized panoramic image data to determine whether the vegetables to be sorted have defects, including:

[0076] The artificial intelligence deep learning model adopts the ATSS algorithm;

[0077] Using an artificial intelligence deep learning model based on the ATSS algorithm to extract features from the panoramic image data, the features including: color, shape, size or texture;

[0078] Based on the extracted features, defective areas in the panoramic image data are identified, and it is determined whether the vegetables to be sorted have defects.

[0079] Specifically, the artificial intelligence deep learning model based on the ATSS algorithm extracts features from the panoramic image data, including visual attributes such as color, texture, shape, and size.

[0080] The ATSS algorithm optimizes the selection of training samples, allowing the model to focus on samples of defects that are difficult to detect, thereby improving the model's ability to recognize different categories of defects.

[0081] In one embodiment, identifying the defective area in the panoramic image data according to the extracted features and determining whether the vegetables to be sorted have defects includes:

[0082] The artificial intelligence deep learning model generates a corresponding confidence score according to the identified defect area in the panoramic image data;

[0083] Based on the confidence score of the defective area, it is determined whether the vegetables to be sorted have defects.

[0084] Specifically, only when the confidence score of the artificial intelligence deep learning model for a certain defect category exceeds a preset threshold, it is determined to be a real defect, where the threshold is set based on experience, testing or specific business needs to balance false positives and negatives.

[0085] Predictions that do not exceed the threshold are considered false positives and can be ignored, manually reviewed, or further analyzed.

[0086] By using confidence thresholds, the false alarm rate and missed alarm rate can be controlled, ensuring that only high-confidence defects are eliminated, reducing unnecessary waste, and significantly improving the overall efficiency and accuracy of the sorting process.

[0087] In one embodiment, if a defect exists, determining the type of the defect and calculating the defect position includes:

[0088] If a defect is detected, a convolutional neural network-based artificial intelligence deep learning model is used to determine the defect category, wherein the defect category includes: defects of the vegetables to be sorted themselves, and / or foreign body defects on the vegetables to be sorted;

[0089] generating a bounding box for each identified defect and determining a location of the defect in the panoramic image data;

[0090] The bounding box is a rectangular area; the position includes: the coordinates of the upper left corner and the lower right corner of the bounding box.

[0091] Specifically, Figure 6 As shown, by extracting features from the panoramic image data and determining which category the defect belongs to, classification algorithms such as support vector machines (SVM), random forests, or deep neural networks can be used to distinguish and identify different types of defects, including but not limited to yellow leaves, rotten leaves, burnt edges, grass, tree leaves, poplar fluff, green caterpillars, aphids, snails, etc.

[0092] A bounding box is then generated for each detected defect area, which is defined by the coordinates of its upper left and lower right corners, thereby accurately locating the defect position, which not only improves the accuracy of defect detection, but also provides a precise target for subsequent rejection operations, ensuring high efficiency and accuracy of the sorting process.

[0093] In one embodiment, the artificial intelligence deep learning model uses a convolutional neural network to extract features from the panoramic image data.

[0094] In one embodiment, when the defective vegetables are moved to the execution position of the high-speed spraying system, the removal operation is performed according to the defect position, including:

[0095] According to the defect position, the center position of the defect is determined, and an initial blowing execution position is obtained;

[0096] According to the preset blowing compensation and blowing width, the initial blowing execution position is adjusted to obtain the final blowing execution position;

[0097] A high-speed spraying system is used to spray the position where the final spraying is performed to remove defective vegetables.

[0098] In implementation, such as Figure 8 and Fig. 9 As shown, according to the defect position, the precise position of the defect in the image is determined, and then the image coordinates are converted into actual physical coordinates on the conveyor belt to obtain the initial blowing execution position;

[0099] Specifically, the X coordinate of the converted image is calculated to be the X coordinate of the high-speed blowing; and the Y coordinate of the converted high-speed blowing is calculated.

[0100] Next, the final blowing execution position is calculated by combining the preset blowing compensation and blowing width;

[0101] like Fig.10 As shown, the high-speed blowing system includes a high-speed blowing structural carrier 11, a high-speed blowing protective box 24 and a high-speed blowing execution carrier. The high-speed blowing execution carrier (4) quickly releases compressed air according to the calculated blowing position to form a high-speed airflow, thereby blowing defective vegetables from the food-grade transparent conveyor belt to the waste area.

[0102] like Figure 1 As shown, the sizes of the waste area and the good product area can be reasonably set according to the movement trajectory of the vegetables when blown down at high speed.

[0103] For example, there is a blowing device at the appropriate position of the flight parabola trajectory. Waste products will be blown, and a new trajectory will be generated due to the flight inertia, blowing force, and gravity. The blowing landing point of the waste products is obtained according to the new trajectory, and the waste area is set; the good product area is set according to the parabola landing point. High-speed transportation can increase the horizontal distance of the flight parabola of the material vegetables from the departure point to the landing point, ensuring that the vegetable waste can be accurately removed in the designated area (waste area), and the good vegetable can also accurately enter the collection area (good product area), thereby achieving efficient separation and collection in the vegetable sorting process.

[0104] In one embodiment, the time from the shooting point to the rejection point is extremely short, less than 1 second, thereby ensuring that the position and state of the material (vegetables) on the high-speed conveyor belt remain unchanged from the time it is shot to the time it reaches the spray landing point, thereby ensuring that the high-speed spraying system can accurately remove defective vegetables from the conveyor belt.

[0105] In one embodiment, the step of using a high-speed spray system to spray the defective position to remove defective vegetables further comprises:

[0106] The spray width is adjusted by a preset coefficient, wherein the coefficient is adjusted according to the size of different vegetables and the type of defects.

[0107] Specifically, according to the size of the vegetables and the severity of the defects, the spraying width is further adjusted by multiplying a coefficient to adapt to different situations and ensure the accuracy and effectiveness of the spraying operation.

[0108] In one embodiment, after the blowing operation, it is confirmed whether the defective vegetables have been successfully removed and feedback is provided. Based on the feedback results, the blowing compensation, blowing width or other parameters are dynamically adjusted to optimize the removal operation.

[0109] Based on the same inventive concept, the present application also provides a high-speed vegetable sorting system, including: a vegetable sorting panoramic shooting system, a defect recognition system and a high-speed blowing system;

[0110] The vegetable sorting panoramic shooting system is used to obtain the front image and the back image of the vegetables to be sorted, and synthesize the front image and the back image into the panoramic image data of the vegetables to be sorted;

[0111] The front image of the vegetables to be sorted is obtained by an upper shooting module arranged above the food-grade transparent conveyor belt; the back image of the vegetables to be sorted is obtained by a lower shooting module arranged below the food-grade transparent conveyor belt;

[0112] The defect recognition system uses an artificial intelligence deep learning model to analyze the panoramic image data to determine whether the vegetables to be sorted have defects;

[0113] If a defect exists, determining the type of the defect and calculating the defect position, wherein the defect position includes the position of the defect in the panoramic image data;

[0114] The high-speed spraying system performs a rejection operation according to the type and position of the defect provided by the defect recognition system when the defective vegetable moves to the execution area of ​​the high-speed spraying system.

[0115] Furthermore, the artificial intelligence deep learning model in the vegetable high-speed sorting system is based on a convolutional neural network and includes:

[0116] Using an artificial intelligence deep learning model based on a convolutional neural network to extract features from the panoramic image data, the features include but are not limited to: color, shape, size or texture;

[0117] Based on the extracted features, defective areas in the panoramic image data are identified, and it is determined whether the vegetables to be sorted have defects.

[0118] Preferably, the high-speed spraying system in the high-speed vegetable sorting system is also used to adjust the coverage of the spraying airflow according to the size of the vegetables and the type of defects.

[0119] This application achieves efficient and accurate sorting of fruits and vegetables by adopting a vegetable sorting panoramic photography system, a high-speed spraying system, and an artificial intelligence deep learning model based on a convolutional neural network, thereby improving the quality score of raw materials, allowing customized defect standards, and supporting a variety of fruit and vegetable sorting processes. It can provide flexible and efficient sorting solutions based on characteristics such as color, size, length, and shape, effectively identify defects of the same color, greatly improve recognition accuracy, increase production capacity, and achieve the flexibility of multiple uses of one machine, support mixed inspection, significantly improve product quality and finished product rate, greatly reduce labor intensity, and enhance the core competitiveness of the enterprise.

[0120] Therefore, the present invention provides a vegetable sorting method and equipment that integrates high-speed conveying technology, high-precision sensors, artificial intelligence technology and a high-speed blowing system. It can take panoramic photos of vegetables, accurately identify various defects, and achieve high-speed rejection and optimized grading, thereby providing customers with efficient and accurate vegetable sorting solutions, greatly improving sorting efficiency and accuracy, reducing production costs, ensuring food safety, and bringing significant economic and social benefits to enterprises.

[0121] In this specification, the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the embodiments described later, the description is relatively simple, and the relevant parts can be referred to the partial description of the previous embodiments.

[0122] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be based on the protection scope of the claims.

Claims

1. A high-speed vegetable sorting method, characterized in that: include: The upper shooting module disposed above the food-grade transparent conveyor belt is used to obtain the front image of the vegetables to be sorted, and the lower shooting module disposed below the food-grade transparent conveyor belt is used to obtain the back image of the vegetables to be sorted; Combining the front image and the back image into panoramic image data of the vegetables to be sorted; Using an artificial intelligence deep learning model to identify the synthesized panoramic image data, and determine whether the vegetables to be sorted have defects; If a defect exists, determining the type of the defect and calculating the defect position, wherein the defect position includes the position of the defect in the panoramic image data; When vegetables with defects move to the execution area of ​​the high-speed spraying system, they are removed according to the type and location of the defects.

2. The high-speed vegetable sorting method according to claim 1, characterized in that: The artificial intelligence deep learning model is used to identify the synthesized panoramic image data to determine whether the vegetables to be sorted have defects, including: The artificial intelligence deep learning model adopts a convolutional neural network; Using an artificial intelligence deep learning model based on a convolutional audit network to extract features from the panoramic image data, the features include: color, shape, size or texture; Based on the extracted features, defective areas in the panoramic image data are identified, and it is determined whether the vegetables to be sorted have defects.

3. The high-speed vegetable sorting method according to claim 2, characterized in that: The step of identifying defective areas in the panoramic image data based on the extracted features and determining whether the vegetables to be sorted have defects includes: The artificial intelligence deep learning model generates a corresponding confidence score according to the identified defect area in the panoramic image data; Based on the confidence score of the defective area, it is determined whether the vegetables to be sorted have defects.

4. The vegetable high-speed sorting method according to claim 1, characterized in that: If a defect exists, the defect type is determined and the defect location is calculated, including: If a defect is detected, a convolutional neural network-based artificial intelligence deep learning model is used to determine the defect category, wherein the defect category includes: defects of the vegetables to be sorted themselves, and / or foreign body defects on the vegetables to be sorted; generating a bounding box for each identified defect and determining a location of the defect in the panoramic image data; The bounding box is a rectangular area; the position includes: the coordinates of the upper left corner and the lower right corner of the bounding box.

5. The high-speed vegetable sorting method according to claim 2, characterized in that: The artificial intelligence deep learning model is based on a convolutional neural network; Features are extracted from the panoramic image data based on a convolutional neural network.

6. The high-speed vegetable sorting method according to claim 1, characterized in that: When the defective vegetables move to the execution position of the high-speed spraying system, a rejection operation is performed according to the defect position, including: According to the defect position, the center position of the defect is determined, and an initial blowing execution position is obtained; According to the preset blowing compensation and blowing width, the initial blowing execution position is adjusted to obtain the final blowing execution position; A high-speed spraying system is used to spray the position where the final spraying is performed to remove defective vegetables.

7. The high-speed vegetable sorting method according to claim 6, characterized in that: The method further comprises: using a high-speed spraying system to spray the defective position to remove the defective vegetables; The spray width is adjusted by a preset coefficient, wherein the coefficient is adjusted according to the size of different vegetables and the type of defects.

8. A high-speed vegetable sorting system, characterized in that: include: Vegetable sorting panoramic photography system, defect recognition system and high-speed spraying system; The vegetable sorting panoramic shooting system is used to obtain the front image and the back image of the vegetables to be sorted, and synthesize the front image and the back image into the panoramic image data of the vegetables to be sorted; The front image of the vegetables to be sorted is obtained by an upper shooting module arranged above the food-grade transparent conveyor belt; the back image of the vegetables to be sorted is obtained by a lower shooting module arranged below the food-grade transparent conveyor belt; The defect recognition system uses an artificial intelligence deep learning model to analyze the panoramic image data to determine whether the vegetables to be sorted have defects; If a defect exists, determining the type of the defect and calculating the defect position, wherein the defect position includes the position of the defect in the panoramic image data; The high-speed spraying system performs a rejection operation according to the type and position of the defect provided by the defect recognition system when the defective vegetable moves to the execution area of ​​the high-speed spraying system.

9. The vegetable high-speed sorting system according to claim 8, characterized in that: The artificial intelligence deep learning model adopts a convolutional neural network, including: Using an artificial intelligence deep learning model based on a convolutional neural network to extract features from the panoramic image data, the features include but are not limited to: color, shape, size or texture; Based on the extracted features, defective areas in the panoramic image data are identified, and it is determined whether the vegetables to be sorted have defects.

10. The vegetable high-speed sorting system according to claim 8, characterized in that: The high-speed spraying system is also used to adjust the coverage of the spraying airflow according to the size of the vegetables and the type of defects.

Images