Identification model training method and device, sorting method and device, equipment and medium

Abstract

The invention provides a recognition model training method and device, a sorting method and device, equipment and a medium, and relates to the technical field of intelligent sorting, and the sorting method comprises the steps: obtaining image data and infrared data of a to-be-recognized potato; using the trained identification model to analyze and process the image data and the infrared data of the to-be-identified potato to obtain an identification target; generating a sorting instruction based on the identification target and the transportation speed; and the sorting mechanism is controlled to execute the sorting instruction to remove sundries and waste potatoes in the potatoes to be recognized, so that the technical problems of poor sorting precision and high cost in the prior art are solved, and the sorting efficiency and the sorting precision are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent sorting, and in particular, to a method for training an identification model, a sorting method, a device, a device and a medium. Background Art

[0002] Currently, for harvested potatoes, manual or intelligent sorting systems are usually used to sort out sundries such as soil and stones and damaged waste potatoes in the potatoes. Among them, for manual sorting, due to the increase in production, not only a large amount of labor is required, but also the labor intensity increases, thereby increasing costs.

[0003] The existing intelligent sorting systems have the following defects: First, the sorting equipment is huge in volume and high in power consumption, making it difficult to be used for on-site sorting and only suitable for use in the workshop; second, the near-infrared detection technology is used for detection and sorting, but this sorting method is inefficient; third, the color selection technology is used for sorting, but problems such as the inability to sort complex and diverse potatoes and the quality inside the potatoes cannot be sorted, etc. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method for training an identification model, a sorting method, a device, a device and a medium to alleviate the technical problems of poor sorting accuracy and high cost existing in the prior art.

[0005] In the first aspect, the present invention provides a method for training an identification model, which is used to identify sundries and waste potatoes in potatoes, and the training method includes: Obtain a training data set; wherein, the training data set includes a plurality of training sample data, and each training sample data includes image data and infrared data of the potato to be identified and the standard identification target in the potato to be identified, and the identification target includes sundries and waste potatoes; Based on the training data set, perform iterative model training operations on the initial identification model until it is determined that the iterative training termination condition is met, and based on the weights and thresholds of the initial identification model updated when the last model training operation is performed, obtain the identification model; Among them, the model training operation includes: selecting target training sample data from the training data set; inputting the image data and infrared data of the potato to be identified in the target training sample data into the initial identification model, so that the initial identification model receives the image data and infrared data of the potato to be identified through the input layer, and processes the image data and infrared data of the potato to be identified through the hidden layer to obtain the identification target in the potato to be identified, and then outputs the identification target corresponding to the potato to be identified through the output layer; based on the prediction error between the identification target and the standard identification target in the potato to be identified in the target training sample data, update the weights and thresholds of the initial identification model.

[0006] In a second aspect, the present invention provides a sorting method, which is applicable to sorting sundries and waste potatoes in potatoes. The method includes: Obtaining image data and infrared data of the potatoes to be identified located within the area of the detection mechanism; Analyzing and processing the image data and infrared data of the potatoes to be identified by using an identification model to obtain an identification target; wherein, the identification model is a model trained by using the above-mentioned identification model training method; Generating a sorting instruction based on the identification target and the transportation speed; Controlling the sorting mechanism to execute the sorting instruction so as to remove the identification target from the potatoes to be identified.

[0007] Optionally, before obtaining the image data and infrared data of the potatoes to be identified located within the area of the detection mechanism, it further includes: Controlling the filtering mechanism to preprocess the original potatoes to be identified so as to remove the soil on the surface of the original potatoes to be identified.

[0008] Optionally, generating a sorting instruction based on the identification target and the transportation speed includes: Determining the center point coordinates of the identification target based on the image data of the potatoes to be identified and the identification target; Determining the coordinates of the identification target on the transmission mechanism based on the center point coordinates of the identification target; Obtaining the sorting coordinates corresponding to the identification target based on the coordinates of the identification target on the transmission mechanism and the transportation speed of the transmission mechanism.

[0009] Optionally, determining the coordinates of the identification target on the transmission mechanism based on the center point coordinates of the identification target includes: Determining the coordinates of the identification target on the transmission mechanism based on the mapping relationship between the coordinate system of the determined image data and the coordinate system of the transmission mechanism and the center point coordinates of the identification target.

[0010] Optionally, obtaining the sorting coordinates based on the coordinates of the identification target on the transmission mechanism and the transportation speed of the transmission mechanism further includes: Obtaining the response time of the sorting mechanism; Determining the action coordinates for the sorting mechanism to sort the identification target based on the sorting coordinates and the response time of the sorting mechanism.

[0011] In a third aspect, the present invention provides a sorting device applicable to the above sorting method. The sorting device includes: a filtering mechanism, a detection mechanism, a transmission mechanism, a sorting mechanism, and a controller; the filtering mechanism is located at one end of the transmission mechanism, the detection mechanism and the sorting mechanism are sequentially located above the transmission mechanism, and the controller is respectively connected to the filtering mechanism, the detection mechanism, the transmission mechanism, and the sorting mechanism; The controller is configured to: control a filtering mechanism to preprocess the original potato to be identified so as to remove the soil on the surface of the original potato to be identified; control a conveying mechanism to move the potato to be identified to the positions of a detection mechanism and a sorting mechanism; control the detection mechanism to obtain image data and infrared data of the potato to be identified within the area of the detection mechanism; analyze and process the image data and infrared data of the potato to be identified by using an identification model to obtain an identification target, wherein the identification model is a model trained by using the above-mentioned identification model training method; generate a sorting instruction based on the identification target and the transportation speed; and control the sorting mechanism to execute the sorting instruction so as to remove the identification target from the potatoes to be identified.

[0012] Optionally, the sorting mechanism includes: air valves, and a plurality of air valves are uniformly arranged along a direction perpendicular to the conveying direction of the conveying mechanism.

[0013] In a fourth aspect, the present invention further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, the above-mentioned identification model training method or the above-mentioned sorting method is implemented.

[0014] In a fifth aspect, the present invention further provides a computer-readable storage medium, which stores computer instructions. When the computer instructions are executed by a processor, the above-mentioned identification model training method or the above-mentioned sorting method is implemented.

[0015] The identification model training method, sorting method, device, equipment, and medium provided by the embodiments of the present invention. The sorting method obtains image data and infrared data of the potato to be identified; analyzes and processes the image data and infrared data by using a trained identification model to obtain an identification target; generates a sorting instruction based on the identification target and the transportation speed; and controls the sorting mechanism to execute the sorting instruction so as to remove the identification target from the potatoes to be identified, which can alleviate the technical problems of poor sorting accuracy and high cost in the prior art, improve the sorting efficiency and sorting accuracy, and can realize the high-speed automatic removal of sundries and waste potatoes during the field harvesting of potatoes. The removal accuracy can reach more than 95%, and at the same time, the labor can be saved by more than 90%.

[0016] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following specific preferred embodiments are given in conjunction with the accompanying drawings and are described in detail as follows. Description of the Drawings

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for use in the embodiments. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0018] Figure 1 Shows a flowchart of a method for training an identification model provided by an embodiment of the present invention; Figure 2 Shows a flowchart of a sorting method provided by an embodiment of the present invention; Figure 3 Shows a schematic structural diagram of a sorting device provided by an embodiment of the present invention; Figure 4 Shows a schematic structural diagram of an electronic device provided by an embodiment of the present invention. Detailed implementation manners

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Usually, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed present invention, but only represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0020] To facilitate better understanding of the present application by those of ordinary skill in the art, the following will first briefly introduce the technical terms involved in the present application.

[0021] Image data is data used to reflect the debris in potatoes, the data of potatoes with external damage, and the data of potatoes mixed with debris. In the present application, the image data can be obtained by shooting potatoes with a high-speed industrial camera to obtain the image data of potatoes.

[0022] Infrared data is data used to reflect potatoes with internal damage. In the present application, the infrared data can be obtained by shooting potatoes with a hyperspectral near-infrared camera to obtain the infrared data of potatoes.

[0023] The recognition target is data used to reflect the presence of impurities and waste potatoes in potatoes. In this application, the recognition target includes impurities such as soil clods and stones, as well as potatoes with deformities, mechanical injuries, green heads, hollowness, black hearts, and disease rot.

[0024] The recognition model is a convolutional neural network model (Convolutional Neural Networks, CNN) that realizes the recognition of potatoes by learning the mapping relationship between the image data and infrared data of potatoes and the impurities and waste potatoes in potatoes; in this application, the recognition model includes an input layer, a hidden layer, and an output layer. The number of input layers and hidden layers is at least one input layer, and each hidden layer includes a convolutional layer, a pooling layer, and a fully connected layer; where: The convolutional layer extracts feature data from the image data and infrared data through a convolutional operation, and its expression is:

[0025] In the formula, is the convolutional kernel, is the input data, is the position of the pixel on the output feature map, is the bias term, is the activation function, is the feature data; The pooling layer reduces the spatial dimension of the feature data through max pooling or average pooling, and its expression is:

[0026] In the formula, is the output data after pooling, is the input data, is the stride, is the position on the output map after pooling; Each neuron in the fully connected layer is connected to all neurons in the previous layer, and its expression is:

[0027] In the formula, is the weight, is the output data of the previous layer, is the bias term of the fully connected layer.

[0028] The "and / or" mentioned in this application describes the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. The character " / " generally represents an "or" relationship between the associated objects before and after.

[0029] After introducing the technical terms involved in this application, next, the technical solutions provided by this application will be described in detail.

[0030] An embodiment of this application provides a method for training an identification model. Refer to Figure 1 As shown, the general process of the method for training an identification model provided by an embodiment of this application is as follows: Step 110: Obtain a training data set; where the training data set includes multiple training sample data, and each training sample data includes image data and infrared data of the potato to be identified, as well as standard identification targets in the potato to be identified, and the identification targets include sundries and waste potatoes.

[0031] In the embodiment of this application, when obtaining the training data set, the following methods can be used but are not limited to: First, obtain the image data and infrared data of the potato to be identified, as well as the infrared identification model, image identification model, and identification targets applicable to the potato to be identified; Then, based on the infrared identification model and the image identification model, process the image data and infrared data of the potato to be identified, and obtain the identification targets corresponding to the potato to be identified during the processing; Finally, based on the image data, infrared data, and identification targets of the potato to be identified, obtain the training data set.

[0032] Furthermore, in the infrared identification model, a partial least squares regression model can be used to determine the starch content and sugar content of the spectrum corresponding to the waste potatoes (such as hollow, black-hearted, and diseased and rotten potatoes) of the potato, and the waste potatoes can be determined with an accuracy rate of up to 98.2%; In the image identification model, for green-headed potatoes, the HSV (Hue, Saturation, Value) color space information can be extracted and the binarization method can be used for determination, with an accuracy rate greater than 99%; for mechanically damaged potatoes, the accumulator method of hough (Hough transform) can be used for determination, with an accuracy rate of up to 97.5%; for deformed potatoes, the potato contour can be extracted by threshold segmentation and the maximum and minimum diameters of the potato can be calculated, or the methods of KNN (k-NearestNeighbors), SVM (Support Vector Machine), and MLP (Multilayer Perceptron) can be used for determination, with an accuracy rate of up to 95%; for sundries such as soil and stones, the YOLO (You Only Look Once) algorithm can be used for determination, with an accuracy rate greater than 99%.

[0033] Step 120: Select target training sample data from the training data set; Step 130: Input the image data and infrared data of the potato to be recognized in the target training sample data into the initial recognition model, so that the initial recognition model receives the image data and infrared data of the potato to be recognized through the input layer, processes the image data and infrared data of the potato to be recognized through the hidden layer to obtain the recognition target in the potato to be recognized, and then outputs the recognition target corresponding to the potato to be recognized through the output layer.

[0034] Step 140: Update the weights and thresholds of the initial recognition model based on the prediction error between the recognition target and the standard recognition target in the potato to be recognized in the target training sample data.

[0035] Step 150: Determine whether the iterative training termination condition is satisfied. If so, execute Step 160; if not, return to Step 120. The iterative training termination condition is that the number of iterations is not less than the iteration number threshold, or the prediction error is not higher than the prediction error threshold.

[0036] Step 160: Obtain the recognition model based on the weights and thresholds of the initial recognition model updated when the model training operation is finally executed.

[0037] Furthermore, in this application, the recognition model is also trained for the recognition of potato varieties, and the accuracy rate of recognizing potato varieties is greater than 90%. Recognizing the varieties of potatoes can play a great role in the preliminary evaluation of potato quality.

[0038] Based on the above embodiments, the embodiments of this application also provide a sorting method. This sorting method sorts the sundries and waste potatoes in the potatoes based on the above-trained recognition model. Refer to Figure 2 As shown, the general process of the sorting method provided by the embodiments of this application is as follows: Step 210: Obtain the image data and infrared data of the potato to be recognized located within the detection mechanism area.

[0039] In the embodiments of this application, before obtaining the image data and infrared data of the potato to be recognized located within the detection mechanism area, it further includes: Control the filtering mechanism to preprocess the original potato to be recognized to remove the soil on the surface of the original potato to be recognized.

[0040] Specifically, the preprocessing of the original potato to be recognized includes, but is not limited to, the following methods: The soil or some soil residues on the surface of the original potatoes to be recognized can be removed manually; or a filtering mechanism can be used to preprocess the original potatoes to be recognized, removing some soil residues while partially removing the soil on the surface of the original potatoes to be recognized, and evenly spreading the processed potatoes to be recognized to avoid stacking; among them, the filtering mechanism can be a cloth cleaning roller.

[0041] In the embodiments of the present application, when acquiring the image data and infrared data of the potatoes to be recognized within the area of the detection mechanism, the following methods can be adopted but are not limited to: Dynamically photograph the potatoes to be recognized on the transmission mechanism through the high-speed industrial camera in the detection mechanism to obtain the image data of the potatoes to be recognized; Photograph the potatoes to be recognized on the transmission mechanism through the hyperspectral near-infrared camera in the detection mechanism to obtain the infrared data of the potatoes to be recognized.

[0042] Step 220: Analyze and process the image data and infrared data of the potatoes to be recognized by using the recognition model to obtain the recognition target; among them, the recognition model is the model trained by the above recognition model training method.

[0043] In the embodiments of the present application, the input data is the image data and infrared data of the potatoes to be recognized; the output data is sundries and waste potatoes (for example, the output data includes recognition targets such as soil blocks, stones, green-headed potatoes, mechanically damaged potatoes, hollow potatoes, black-hearted potatoes, and diseased and rotten potatoes, etc.); after the recognition module receives the image data and infrared data of the potatoes to be recognized through the input layer, processes the image data and infrared data of the potatoes to be recognized through the hidden layer, and then outputs the recognition target and the relative coordinate data of the recognition target in the image data through the output layer. In this way, by using the recognition model with the characteristics of the convolutional neural network, based on the image data and infrared data of the potatoes, predicting the sundries and waste potatoes in the potatoes to be recognized can improve the sorting efficiency and sorting accuracy of the potatoes, and the sorting accuracy can reach more than 95%, while saving more than 90% of the labor.

[0044] Step 230: Generate a sorting instruction based on the recognition target and the transportation speed; In the embodiments of the present application, when generating the sorting instruction, the following methods can be adopted but are not limited to: First, based on the image data of the potatoes to be recognized and the recognition target, determine the center point coordinates of the recognition target; Secondly, based on the center point coordinates of the recognition target, determine the coordinates of the recognition target on the transmission mechanism; among them, based on determining the mapping relationship between the coordinate system of the image data and the coordinate system of the transmission mechanism and the center point coordinates of the recognition target, determine the coordinates of the recognition target on the transmission mechanism; Then, based on the coordinates of the recognition target on the conveying mechanism and the conveying speed of the conveying mechanism, the sorting coordinates corresponding to the recognition target are obtained. Among them, the response time of the sorting mechanism can be obtained; based on the sorting coordinates and the response time of the sorting mechanism, the action coordinates for the sorting mechanism to sort the recognition target are determined.

[0045] Specifically, first, the recognition target in the image data of the potato to be recognized can be determined through the above steps, that is, the sundries and waste potatoes to be removed in each image; based on the determined recognition target and the image data, the center point coordinates of the recognition target can be determined, and its center point coordinates are:

[0046] In the formula, and are the coordinates of the bounding box of the recognition target in the image data.

[0047] Then, based on the mapping relationship established between the image coordinate system (pixels) and the conveying mechanism coordinate system (millimeters), the coordinates of the recognition target on the conveying mechanism are determined. Among them, the mapping relationship is:

[0048] In the formula, and are both coefficients, generally determined through calibration experiments and The coefficient values of, in this application, , .

[0049] Since the conveying mechanism only moves in one direction, therefore, in the embodiment of this application, the mapping relationship between the image coordinate system and the conveying mechanism coordinate system is the mapping relationship in the conveying direction of the conveying mechanism, that is, the X-axis coordinate. In the subsequent process of calculating coordinates, only the coordinate values in the X-axis direction are processed.

[0050] Finally, according to the conveying speed of the conveying mechanism and the response time of the sorting mechanism, the coordinates for the sorting mechanism to perform the sorting operation are determined, that is, the working lead of the sorting mechanism. The coordinates are:

[0051] In the formula, is the speed of the conveying mechanism, with the unit of mm / s, is the response time of the sorting mechanism, with the unit of s.

[0052] Based on the coordinates for the sorting mechanism to perform the sorting operation and the coordinates of the recognition target on the conveying mechanism, the action coordinates for the sorting mechanism to sort the recognition target are determined. The action coordinates are:

[0053] Further, based on the action coordinates of the sorting mechanism for sorting and identifying the target and the control strategy of the sorting mechanism, a sorting instruction is generated.

[0054] Step 240: Control the sorting mechanism to execute the sorting instruction to remove the identified target from the potatoes to be identified.

[0055] In the embodiment of the present application, the sorting mechanism can be controlled to execute the sorting instruction to remove sundries and waste potatoes from the potatoes. The sorting mechanism can adopt a pneumatic blowing mechanism or a flicking mechanism or other mechanisms that can be used for sorting.

[0056] Further, by the sorting mechanism executing the sorting instruction, the sundries and waste potatoes in the potatoes to be identified can be moved to the removal conveyor mechanism, and at the same time, the conveyor mechanism moves the target potatoes to the discharge conveyor mechanism, thereby completing the sorting of the sundries and waste potatoes in the potatoes.

[0057] In the embodiment of the present application, based on the multi-modal convolutional neural network to identify the biological characteristics and quality of potatoes, the method for identifying the grade of potatoes through dual-source identification can accurately and efficiently remove sundries and waste potatoes from the potatoes.

[0058] Based on the above embodiments, the embodiment of the present application provides a sorting device. Refer to Figure 3 As shown, the sorting device provided by the embodiment of the present application at least includes: a filtering mechanism 310, a detection mechanism 330, a conveyor mechanism 320, a sorting mechanism 340, and a controller; the filtering mechanism 310 is located at one end of the conveyor mechanism 320, the detection mechanism 330 and the sorting mechanism 340 are sequentially located above the conveyor mechanism, and the controller is respectively connected to the filtering mechanism 310, the detection mechanism 330, the conveyor mechanism 320, and the sorting mechanism 340; The controller is configured to: control the filtering mechanism 310 to preprocess the original potatoes to be identified to remove the soil on the surface of the original potatoes to be identified; control the conveyor mechanism 320 to move the potatoes to be identified to the positions of the detection mechanism 330 and the sorting mechanism 340; control the detection mechanism 330 to obtain the image data and infrared data of the potatoes to be identified located within the area of the detection mechanism; analyze and process the image data and infrared data of the potatoes to be identified by using an identification model to obtain an identified target; wherein, the identification model is a model trained by using the above identification model training method; generate a sorting instruction based on the identified target and the transportation speed; control the sorting mechanism 340 to execute the sorting instruction to remove the identified target from the potatoes to be identified.

[0059] In the embodiment of the present application, the sorting mechanism includes: air valves, and a plurality of air valves are uniformly arranged along the conveying direction perpendicular to the conveyor mechanism.

[0060] In an embodiment of the present application, the sorting device further includes: a rejection conveying mechanism 360, a discharging conveying mechanism 350, and an earth removal mechanism 370. Among them, the rejection conveying mechanism 360 and the discharging conveying mechanism 350 are arranged at one end of the conveying mechanism away from the detection mechanism. The rejection conveying mechanism 360 is used to move sundries and waste potatoes to a position for storing sundries and waste potatoes, and the discharging conveying mechanism 350 is used to move target potatoes to a designated position; the earth removal mechanism 370 is arranged below the filtering mechanism 310 and is used to convey the soil in the original potatoes to a designated position.

[0061] In an embodiment of the present application, during the process of removing sundries and waste potatoes from the potatoes to be identified by the sorting device, the working process of the controller is as follows: First, the potatoes automatically harvested by the combine harvester are conveyed to the filtering mechanism. The controller controls the filtering mechanism to preprocess the original potatoes to be identified to obtain the potatoes to be identified, and conveys the potatoes to be identified to the initial position of the conveying mechanism; Secondly, control the conveying mechanism to move the potatoes to be identified to the position of the detection mechanism, and control the detection mechanism to obtain the image data and infrared data of the potatoes to be identified; Then, the controller uses the recognition model to analyze and process the image data and infrared data of the potatoes to be identified to obtain the recognition target; based on the recognition target and the transportation speed, generate a sorting instruction; Finally, the controller controls the jet air valve of the sorting mechanism to blow air at the recognition target, so that sundries or waste potatoes are conveyed to the rejection conveying mechanism, and the conveying mechanism conveys the target potatoes to a designated position.

[0062] It should be noted that the principle of the sorting device provided in the embodiment of the present application for solving technical problems is similar to the sorting method provided in the embodiment of the present application. Therefore, for the implementation of the sorting device provided in the embodiment of the present application, reference can be made to the implementation of the sorting method provided in the embodiment of the present application, and repeated parts will not be elaborated.

[0063] After introducing the sorting method and device provided in the embodiment of the present application, next, a brief introduction to the electronic device provided in the embodiment of the present application will be given.

[0064] Refer to Figure 4 As shown, the electronic device 500 provided in the embodiment of the present application at least includes a processor 501, a memory 502, and a computer program stored on the memory 502 and executable on the processor 501. When the processor 501 executes the computer program, the sorting method provided in the embodiment of the present application is implemented.

[0065] The electronic device 500 provided by the embodiments of the present application may further include a bus 503 connecting different components (including a processor 501 and a memory 502). Among them, the bus 503 represents one or more of several types of bus structures, including a memory bus, a peripheral bus, a local bus, etc.

[0066] The memory 502 may include a readable storage medium in the form of a volatile memory, such as a random access memory (RAM) 5021 and / or a cache memory 5022, and may further include a read-only memory (ROM) 5023. The memory 502 may also include a program tool 5025 having a set (at least one) of program modules 5024. The program modules 5024 include, but are not limited to, an operating subsystem, one or more application programs, other program modules, and program data. Each or some combination of these examples may include the implementation of a network environment.

[0067] The processor 501 may be a processing element or a collective term for multiple processing elements. For example, the processor 501 may be a central processing unit (CPU), or one or more integrated circuits configured to implement the rocket engine stability prediction method provided by the embodiments of the present application. Specifically, the processor 501 may be a general-purpose processor, including but not limited to a CPU, an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0068] The electronic device 500 may communicate with one or more external devices 504 (such as a keyboard, a remote control, a sensor, a camera, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 500 (such as a mobile phone, a computer, etc.), and / or communicate with a device that enables the electronic device 500 to communicate with one or more other electronic devices 500 (such as a router, a modem, etc.). Such communication may be performed through an input / output (I / O) interface 505. And, the electronic device 500 may also communicate with one or more networks (such as a local area network (LAN), a wide area network (WAN), and / or a public network, such as the Internet) through a network adapter 506. As Figure 4As shown, network adapter 506 communicates with other modules of electronic device 500 via bus 503. It should be understood that although Figure 4 not shown in Figure 4 , other hardware and / or software modules may be used in conjunction with electronic device 500, including but not limited to microcode, device drivers, redundant processors, external disk drive arrays, redundant arrays of independent disks (RAID) subsystems, tape drives, and data backup storage subsystems, etc.

[0069] It should be noted that Figure 4 the electronic device 500 shown is merely an example and should not impose any limitations on the functions and usage scope of the embodiments of the present application.

[0070] Next, the computer-readable storage medium provided by the embodiments of the present application will be introduced. The computer-readable storage medium provided by the embodiments of the present application stores computer instructions, and when these computer instructions are executed by a processor, the rocket engine stability prediction method provided by the embodiments of the present application is implemented. Specifically, these computer instructions can be built-in or installed in the processor, so that the processor can implement the sorting method provided by the embodiments of the present application by executing the built-in or installed computer instructions.

[0071] In addition, the rocket engine stability prediction method provided by the embodiments of the present application can also be implemented as a computer program product. This computer program product includes program code, and when this program code runs on a processor, the sorting method provided by the embodiments of the present application is implemented.

[0072] The computer program product provided by the embodiments of the present application may employ one or more computer-readable storage media, and a computer-readable storage medium may be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the above. Specifically, more specific examples (non-exhaustive list) of computer-readable storage media include electrical connections with one or more wires, portable disks, hard disks, RAM, ROM, erasable programmable read-only memory (EPROM), optical fibers, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0073] The computer program product provided by the embodiments of the present application may be a CD-ROM and include program codes, and may also run on an electronic device such as a computer. However, the computer program product provided by the embodiments of the present application is not limited thereto. In the embodiments of the present application, a computer-readable storage medium may be any tangible medium that contains or stores program codes, and the program codes may be used by or in combination with an instruction execution system, apparatus, or device.

[0074] It should be noted that although several units or subunits of the apparatus are mentioned in the above detailed description, this division is merely exemplary and not mandatory. In fact, according to the embodiments of the present application, the features and functions of the two or more units described above may be embodied in one unit. Conversely, the features and functions of one unit described above may be further divided and embodied by multiple units.

[0075] In addition, although the operations of the method of the present application are described in a specific order in the drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, some steps may be omitted, multiple steps may be combined into one step for execution, and / or one step may be decomposed into multiple steps for execution.

[0076] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present application.

[0077] Obviously, those skilled in the art can make various changes and modifications to the embodiments of the present application without departing from the spirit and scope of the embodiments of the present application. Thus, if these modifications and variations of the embodiments of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these changes and modifications.

Claims

1. A recognition model training method, characterized in that: The recognition model is used to identify impurities and waste potatoes in potatoes, and the training method comprises: Acquire a training data set; wherein the training data set includes a plurality of training sample data, each of the training sample data includes image data and infrared data of a potato to be identified and a standard identification target in the potato to be identified, wherein the identification target includes sundries and waste potatoes; Based on the training data set, iteratively perform a model training operation on the initial recognition model until it is determined that an iterative training termination condition is met, and obtain a recognition model based on each weight and each threshold of the initial recognition model updated when the model training operation was last performed; The model training operation includes: selecting target training sample data from the training data set; inputting the image data and infrared data of the potato to be identified in the target training sample data into the initial recognition model, so that the initial recognition model receives the image data and infrared data of the potato to be identified through the input layer, and processes the image data and infrared data of the potato to be identified through the hidden layer to obtain the identification target in the potato to be identified, and then outputs the identification target corresponding to the potato to be identified through the output layer; based on the prediction error between the identification target and the standard identification target in the potato to be identified in the target training sample data, updating the weights and thresholds of the initial recognition model.

2. A sorting method, characterized in that: The method is suitable for sorting sundries and waste potatoes from potatoes, and the method comprises: Acquiring image data and infrared data of potatoes to be identified within the detection mechanism area; The image data and the infrared data of the potato to be identified are analyzed and processed by using a recognition model to obtain an identification target; wherein the recognition model is a model trained by using the recognition model training method according to claim 1; generating sorting instructions based on the identified target and the transport speed; The sorting mechanism is controlled to execute the sorting instruction to remove the identification target from the potatoes to be identified.

3. The sorting method according to claim 2, characterized in that: Before acquiring the image data and infrared data of the potatoes to be identified in the detection mechanism area, the method further includes: The filtering mechanism is controlled to pre-treat the original potatoes to be identified so as to remove the dirt on the surface of the original potatoes to be identified.

4. The sorting method according to claim 2, characterized in that: Based on the identified target and the transport speed, a sorting instruction is generated, including: Based on the image data of the potato to be identified and the identification target, determining the coordinates of the center point of the identification target; Based on the coordinates of the center point of the identified target, determining the coordinates of the identified target at the transmission mechanism; Based on the coordinates of the identified target located on the transmission mechanism and the transport speed of the transmission mechanism, the sorting coordinates corresponding to the identified target are obtained.

5. The sorting method according to claim 4, characterized in that: The coordinates of the center point of the identified target are determined to determine the coordinates of the identified target located at the transmission mechanism, including: Based on determining a mapping relationship between a coordinate system of the image data and a coordinate system of the transmission mechanism and the center point coordinates of the recognition target, the coordinates of the recognition target located at the transmission mechanism are determined.

6. The sorting method according to claim 4, characterized in that: Based on the coordinates of the identified target located at the transmission mechanism and the transport speed of the transmission mechanism, the sorting coordinates are obtained, and the method further includes: Get the response time of the sorting agency; Based on the sorting coordinates and the response time of the sorting mechanism, the action coordinates of the sorting mechanism for sorting the identification target are determined.

7. A sorting device suitable for the sorting method according to any one of claims 2 to 6, characterized in that: The sorting device comprises: a filtering mechanism, a detection mechanism, a transmission mechanism, a sorting mechanism and a controller; the filtering mechanism is located at one end of the transmission mechanism, the detection mechanism and the sorting mechanism are located above the transmission mechanism in sequence, and the controller is connected to the filtering mechanism, the detection mechanism, the transmission mechanism and the sorting mechanism respectively; The controller is used to: control the filtering mechanism to pre-process the original potatoes to be identified so as to remove the dirt on the surface of the original potatoes to be identified; control the transmission mechanism to move the potatoes to be identified to the position of the detection mechanism and the position of the sorting mechanism; control the detection mechanism to obtain the image data and infrared data of the potatoes to be identified in the detection mechanism area; use the recognition model to analyze and process the image data and the infrared data of the potatoes to be identified to obtain the recognition target; wherein the recognition model is a model trained by the recognition model training method according to claim 1; generate a sorting instruction based on the recognition target and the transportation speed; control the sorting mechanism to execute the sorting instruction to remove the recognition target from the potatoes to be identified.

8. The sorting device according to claim 7, characterized in that: The sorting mechanism comprises: a gas valve, and a plurality of the gas valves are evenly arranged along a conveying direction perpendicular to the transmission mechanism.

9. An electronic device, characterized in that: It includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the recognition model training method as described in claim 1 or implements the sorting method as described in any one of claims 2-6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, which, when executed by a processor, implement the recognition model training method as described in claim 1 or the sorting method as described in any one of claims 2-6.

Images