Solid waste identification method, device and equipment and storage medium

Abstract

The invention discloses a solid waste identification method, device and equipment and a storage medium, and relates to the technical field of waste identification, and the solid waste identification method comprises the steps: obtaining element detection data corresponding to a to-be-identified waste; the element detection data is input into a solid waste recognition model, a solid waste recognition result output by the solid waste recognition model is obtained, and the solid waste recognition model is obtained through iterative training based on first element data of a plurality of sets of non-solid waste bodies; and based on the solid waste identification result, determining a target waste category of the to-be-identified waste. According to the method, large-model data analysis is performed on the element detection data corresponding to the to-be-identified waste, so that the labor and time cost consumed by solid waste identification is reduced, the detection period is shortened, and rapid detection is realized.

Description

Technical Field

[0001] The present application relates to the field of waste identification technology, and in particular to a solid waste identification method, device, equipment and storage medium. Background Art

[0002] In various fields such as industrial production, daily life, and construction, a large amount of solid waste is generated. These wastes include but are not limited to metal waste, plastic products, paper, glass, organic waste, etc. Each type of solid waste has its own unique physical and chemical properties. For example, plastic products have different melting points and heat resistance depending on their polymer types and additives. But even for the same type of solid waste, they will show completely different physical and chemical properties due to different sources, production processes, usage history, exposure environment, and degree of pollution, which further increases the difficulty of solid waste identification.

[0003] However, traditional chemical detection methods for hazardous solid waste usually require professional laboratory equipment and professional technicians, are costly, have long detection cycles, and are difficult to achieve rapid on-site detection. Summary of the invention

[0004] The main purpose of this application is to provide a solid waste identification method, device, equipment and storage medium, aiming to solve the problem of high cost and long time in chemical detection of solid waste.

[0005] To achieve the above objectives, the present application proposes a solid waste identification method, the method comprising:

[0006] Obtaining element detection data corresponding to the waste to be identified;

[0007] Inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on a plurality of groups of first element data of non-solid waste objects;

[0008] Based on the solid waste identification result, a target waste category of the waste to be identified is determined.

[0009] In one embodiment, the training process of the solid waste identification model includes:

[0010] Obtaining first element data of several groups of non-solid waste objects;

[0011] Performing random missing processing on each of the first element data to obtain several groups of second element data;

[0012] Each of the second element data is input into an initial solid waste identification model for iterative training to obtain the solid waste identification model.

[0013] In one embodiment, performing random missing processing on each of the first element data to obtain a plurality of sets of second element data includes:

[0014] Obtaining a preset missing element and an average value of missing elements corresponding to the preset missing element;

[0015] The first element data corresponding to the preset missing elements are randomly eliminated, and the missing element average values ​​are used for data filling to obtain several groups of second element data.

[0016] In one embodiment, the initial solid waste identification model includes a data vectorization layer, a feature extraction layer, and a classification layer;

[0017] The step of inputting each of the second element data into an initial solid waste identification model for iterative training to obtain the solid waste identification model comprises:

[0018] Input each of the second element data into the data vectorization layer to obtain a plurality of groups of third element data output by the data vectorization layer;

[0019] Extracting element data features in each of the third element data through the feature extraction layer;

[0020] Inputting the element data features into the classification layer for classification, and obtaining a first prediction value of each object category output by the classification layer;

[0021] Based on the first prediction value, the solid waste identification model is obtained.

[0022] In one embodiment, inputting the element data feature into the classification layer for classification to obtain a first prediction value of each object category output by the classification layer includes:

[0023] Inputting the element data feature into the fully connected layer in the classification layer for classification to obtain a second prediction value;

[0024] Performing category mapping on the second predicted value to obtain a probability distribution value of each object category;

[0025] The probability distribution values ​​are integrated to generate the first prediction value.

[0026] In one embodiment, obtaining the solid waste identification model based on the first prediction value includes:

[0027] Calculating the first prediction value, the preset sample label value, and the preset weight matrix to obtain an error value;

[0028] Comparing the error value with a preset expected range;

[0029] If the error value does not meet the preset expected range, the initial solid waste identification model is iteratively trained based on the error value to obtain the solid waste identification model.

[0030] In one embodiment, after comparing the error value with a preset expected range, the method further includes:

[0031] If the error value meets the preset expected range, the initial solid waste identification model is used as the solid waste identification model;

[0032] Obtaining the prediction accuracy and actual sample number corresponding to the solid waste identification model;

[0033] Based on the prediction accuracy and the actual number of samples, a confusion matrix diagram and a statistical box plot are generated to optimize the solid waste identification model.

[0034] In addition, to achieve the above-mentioned purpose, the present application also proposes a solid waste identification device, the solid waste identification device comprising:

[0035] An acquisition module, used to acquire element detection data corresponding to the waste to be identified;

[0036] An output module, used for inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on several sets of conventional element data;

[0037] A classification module is used to determine the target waste category of the waste to be identified based on the solid waste identification result.

[0038] In addition, to achieve the above-mentioned purpose, the present application also proposes a solid waste identification device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the solid waste identification method described above.

[0039] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by the processor, the steps of the solid waste identification method described above are implemented.

[0040] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the solid waste identification method described above.

[0041] The present application provides a solid waste identification method, device, equipment and storage medium. The solid waste identification method obtains element detection data corresponding to the waste to be identified, and then inputs the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on the first element data of several groups of non-solid waste objects, and then based on the solid waste identification result, the target waste category of the waste to be identified is determined, thereby reducing the labor and time costs consumed by solid waste identification, shortening the detection cycle, and thus achieving rapid detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0044] Figure 1 A schematic diagram of a process flow provided for the first embodiment of the solid waste identification method of the present application;

[0045] Figure 2 A schematic diagram of a flow chart provided for the second embodiment of the solid waste identification method of the present application;

[0046] Figure 3 A schematic diagram of a process flow provided for the third embodiment of the solid waste identification method of the present application;

[0047] Figure 4 A simplified flow chart of the solid waste identification method provided for this application;

[0048] Figure 5 This is a schematic diagram of the module structure of the solid waste identification device according to an embodiment of the present application;

[0049] Figure 6 Schematic diagram of the device structure of the hardware operating environment involved in the solid waste identification method in the embodiment of the present application.

[0050] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0051] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0052] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0053] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a big data service platform, a solid waste identification system, etc. The following takes the solid waste identification system as an example to illustrate this embodiment and the following embodiments.

[0054] Based on this, the present application embodiment provides a solid waste identification method, referring to Figure 1 , Figure 1 A schematic diagram of the process flow provided for Example 1 of the solid waste identification method of the present application.

[0055] In this embodiment, the solid waste identification method includes steps S11 to S13:

[0056] Step S11, obtaining element detection data corresponding to the waste to be identified;

[0057] It should be noted that the waste to be identified refers to waste samples whose specific categories need to be determined, that is, they have not yet been classified and need to be determined through testing and analysis to determine whether they are solid waste, and if so, which category of solid waste the waste to be identified belongs to.

[0058] It should be further explained that the element detection data refers to the data of elements contained in the waste sample, including information such as the type, content, and proportion of the elements, such as 'Fe', 'SiO2', 'CaO', 'Al2O3', 'MgO', 'TiO2', 'MnO', 'P', 'S', and 'FeO', etc., which are used to characterize the chemical composition of the waste.

[0059] Specifically, element analysis is performed on the waste to be identified to obtain its chemical composition data, i.e., element detection data, which can be achieved by spectral analysis, near-infrared spectral analysis and other technologies, which are not limited here, to determine the content of various elements in the waste to be identified.

[0060] Step S12, inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on a plurality of groups of first element data of non-solid waste objects;

[0061] It should be noted that the solid waste identification model is used to identify and classify solid waste based on the input element detection data. The solid waste identification result refers to the result output by the solid waste identification model after analyzing the waste sample to be identified, including the classification label of the waste, such as metal, plastic, glass, etc., and possible confidence or probability scores.

[0062] It should be further explained that the non-solid waste object refers to a substance or material that does not belong to the solid waste category. The first element data refers to the element data corresponding to the non-solid waste object, which is used for iterative training of the solid waste recognition model.

[0063] Step S13: determining the target waste category of the waste to be identified based on the solid waste identification result.

[0064] It should be noted that the target waste category refers to the specific classification or group to which the waste to be identified belongs, such as metal waste, glass waste, etc., determined after model analysis and judgment during the solid waste identification process. There is no restriction here and it can be set according to actual conditions.

[0065] Specifically, based on the solid waste identification result, the target waste category of the waste to be identified is determined, that is, based on the solid waste identification result, it is judged whether the waste to be identified is solid waste, thereby achieving rapid detection of solid waste.

[0066] Furthermore, if the waste to be identified is not solid waste, the possible target waste category of the waste to be identified is output according to the solid waste identification result, so that the decision maker can classify the waste and treat the solid waste according to the target waste category.

[0067] This embodiment obtains element detection data corresponding to the waste to be identified, and then inputs the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on several groups of first element data of non-solid waste objects, and then based on the solid waste identification result, the target waste category of the waste to be identified is determined, thereby reducing the labor and time costs consumed in solid waste identification, shortening the detection cycle, thereby achieving rapid detection, and improving the efficiency and speed of waste treatment.

[0068] Based on this, the present application embodiment provides a solid waste identification method, referring to Figure 2 , Figure 2 A schematic diagram of the process flow provided for Example 2 of the solid waste identification method of the present application.

[0069] In a feasible implementation manner, the training process of the solid waste identification model includes:

[0070] Step S21, obtaining first element data of several groups of non-solid waste objects;

[0071] Specifically, common non-solid waste objects in smelting waste can be obtained through online data query and practical testing of research projects, and then the first element data of these non-solid waste objects can be obtained through data and testing, such as an array containing elements such as 'Fe', 'SiO2', 'CaO', 'Al2O3', 'MgO', 'TiO2', 'MnO', 'P', 'S', 'FeO'. For example, conventional steel generally contains the elements ['Fe', 'MnO', 'P', 'S', 'FeO']. There is no restriction on the acquisition method here, and it can be set according to the actual situation, depending on the classification requirements of solid waste.

[0072] It can be understood that solid waste is identified through the overfitting characteristics of the model to conventional data (i.e., the first element data of non-solid waste objects), that is, after the model learns the characteristics of conventional data, it can accurately identify solid waste. Specifically, conventional data refers to normal, non-abnormal data samples that do not contain solid waste, so that conventional data is used to train deep learning models so that it can learn the characteristics of normal states. When the model encounters input data that is significantly different from these conventional data, such as data containing solid waste, the model will show an abnormal reaction because the characteristics of the solid waste are inconsistent with the characteristics of the conventional data learned by the model. Solid waste is thus identified and distinguished through this abnormal reaction because the introduction of solid waste breaks the normal pattern represented by conventional data.

[0073] Therefore, conventional data is the benchmark for training and calibrating the model, which helps the model distinguish between normal and abnormal states (such as the presence of solid waste). Therefore, there is no need for a large number of solid waste samples to train the model. Instead, the existing conventional data model is used to identify potential solid waste samples through sensitive detection of abnormal situations, thereby realizing an effective and resource-saving solid waste judgment method.

[0074] Step S22, performing random missing processing on each of the first element data to obtain a plurality of groups of second element data;

[0075] It should be noted that the second element data refers to a data set that has been processed with random missing data.

[0076] Specifically, a preset missing element and an average value of missing elements corresponding to the preset missing element are obtained, and then the first element data corresponding to the preset missing element are randomly removed, and the average value of missing elements is used for data filling to obtain several groups of second element data.

[0077] Additionally, each of the second element data can be divided into a training set and a validation set in a ratio of 8:2, that is, 80% of the data is used to train the model so that it can learn and identify the features of the element data, and the remaining 20% ​​of the data is used to verify the performance of the model to ensure that the model has good generalization capabilities.

[0078] Step S23, inputting each of the second element data into an initial solid waste identification model for iterative training to obtain the solid waste identification model.

[0079] It should be noted that the initial solid waste identification model refers to a basic machine learning or deep learning model before starting the training process. The initial model has not been trained, or it has some basic structure and parameter settings, but has not been iteratively trained with actual data.

[0080] Specifically, each of the second element data is input into the data vectorization layer to obtain several groups of third element data output by the data vectorization layer, and then the element data features in each of the third element data are extracted through the feature extraction layer, and then the element data features are input into the classification layer for classification to obtain the first prediction value of each object category output by the classification layer, and then based on the first prediction value, the solid waste recognition model is obtained.

[0081] This embodiment obtains several groups of first element data of non-solid waste objects, and then performs random missing processing on each of the first element data to obtain several groups of second element data, and then inputs each of the second element data into the initial solid waste recognition model for iterative training to obtain the solid waste recognition model, and then simulates the data missing situation in the real world through random missing processing, so that the model can learn how to make accurate predictions when faced with incomplete data, thereby improving the robustness of the model, and at the same time enabling the model to better adapt when faced with new and unseen data, reducing the risk of overfitting, and improving the generalization ability and adaptability of the model, so as to more accurately identify and classify solid waste and improve recognition accuracy.

[0082] In a feasible implementation manner, performing random missing processing on each of the first element data to obtain several groups of second element data includes:

[0083] Step S31, obtaining a preset missing element and an average value of missing elements corresponding to the preset missing element;

[0084] Step S32, randomly remove the first element data corresponding to the preset missing element, and fill the data with the average value of the missing element to obtain several groups of second element data.

[0085] It should be noted that the preset missing elements refer to the elements (features) that are predetermined to be missing in the data set during the data preprocessing stage based on prior knowledge and understanding of the data set. The elements may be data points that are missing due to measurement errors, incomplete data records, equipment failure or other reasons. The identification of preset missing elements helps to prepare processing strategies in advance, such as deciding which features are important in the data set and preparing appropriate processing methods to deal with their absence.

[0086] It should be further explained that the missing element average value refers to the average value of the preset missing elements when they appear in the data set, so that the average value is used as the filling value. When these elements are missing in the actual data, their average value can be used to fill in to maintain the integrity of the data. For example, if the "humidity" feature in a data set is a preset missing element, then the missing element average value of "humidity" is calculated, and then when the "humidity" data is missing, the missing element average value is used to replace it.

[0087] Specifically, in order to solve the problem of reduced model accuracy caused by uncertain missing elements, random missing input data is introduced, that is, when training the model, a random element in the training input data set (each of the first element data) is filled with the average value, and elements are randomly selected to be missing during the training process, that is, the first element data corresponding to the preset missing elements are randomly eliminated, and the missing element average value is used to fill the data to obtain several groups of second element data. In this way, by determining fixed elements and possible missing elements, the data missing that may occur in the real world is simulated, thereby making the model more accurate.

[0088] This embodiment obtains preset missing elements and the average value of missing elements corresponding to the preset missing elements, and then randomly eliminates the first element data corresponding to the preset missing elements, and fills the missing element average value with data to obtain several groups of second element data. By simulating data missing and filling, the model learns how to respond when encountering incomplete data in actual applications, thereby improving the robustness of the model to data missing and the generalization ability of the model when facing unknown data. At the same time, random elimination and filling can reduce the deviation caused by missing data, ensure that model training will not be affected by specific patterns, and thus improve prediction accuracy.

[0089] Based on this, the present application embodiment provides a solid waste identification method, referring to Figure 3 , Figure 3 A schematic diagram of the process flow provided for Example 3 of the solid waste identification method of the present application.

[0090] In a feasible implementation manner, the initial solid waste identification model includes a data vectorization layer, a feature extraction layer, and a classification layer; the inputting each of the second element data into the initial solid waste identification model for iterative training to obtain the solid waste identification model includes:

[0091] Step S41, inputting each of the second element data into the data vectorization layer to obtain a plurality of groups of third element data output by the data vectorization layer;

[0092] It should be noted that the data vectorization layer refers to a layer in the deep learning model that is used to convert structured data into feature vectors. The feature extraction layer refers to the part of the model responsible for extracting useful features from the input data, which can be achieved through convolutional layers, pooling layers or other types of network layers to identify patterns and features in the input data.

[0093] It should be further explained that the classification layer refers to the part of the model responsible for mapping the extracted features to the final output categories. In a typical classification task, the classification layer can be composed of a fully connected layer (also called a dense layer) and an output layer, where the output layer usually uses a softmax function or other activation function to generate a category probability distribution. The third element data refers to the data output after being processed by the data vectorization layer.

[0094] Specifically, each of the second element data is input into the data vectorization layer to obtain several groups of third element data output by the data vectorization layer. After this embedded vectorization, discrete data can be represented by continuous space. In one embodiment, in order to solve the problem of insufficient vectorization and difficulty in information fitting due to the number of feature numbers, 32 vectorized feature numbers are used.

[0095] Step S42, extracting element data features in each of the third element data through the feature extraction layer;

[0096] It should be noted that the element data features refer to useful information extracted from the third element data, which can represent the key attributes and patterns of the data. In the deep learning model, the feature extraction layer automatically identifies and extracts these features for classification and other tasks.

[0097] Specifically, in one embodiment, the feature extraction layer adopts a one-dimensional residual connection convolutional neural network, inputs 32 vectorized features, first passes through input channel 1, convolution layer 1 with a convolution kernel size of 3×3 and a stride of 3, and the output channel is 16; then passes through a 3×3 maximum downsampling pooling layer, and the output channel is 16; then stacks 4 residual network blocks, and the output channel is 512; for each residual network block, the staggered network layer 1: the input channel is 16, first passes through a convolution kernel size of 3×3 and a maximum downsampling pooling operation with a stride of 2; then stacks 3 [1×1, 16; 3×3, 16; 1×1, 64] structured Bottleneck. The output channel of this layer is 64. Staggered network layer 2, the input channel is 64. First, a maximum downsampling pooling operation with a convolution kernel size of 3×3 and a stride of 2 is performed; then three Bottleneck structures of [1×1, 64; 3×3, 64; 1×1, 256] are stacked. The output channel of this layer is 256. Staggered network layer 3: The input channel is 256. This layer stacks 4 Bottleneck structures of [1×1, 128; 3×3, 128; 1×1, 512]. The output channel of this layer is 512.

[0098] Among them, Bottleneck refers to a specific network layer structure used to reduce the number of parameters and the amount of calculation while maintaining network performance. The Bottleneck structure usually contains three convolutional layers: the first 1x1 convolutional layer is used to reduce the number of input channels and the number of parameters. It is followed by a 3x3 convolutional layer for feature extraction. The last 1x1 convolutional layer is used to increase the number of output channels and provide more feature maps for the next layer. The structure reduces the computational complexity and the number of parameters while maintaining the network depth by reducing the dimension of the intermediate feature map (ie, the "bottleneck").

[0099] Step S43, inputting the element data feature into the classification layer for classification, and obtaining a first prediction value of each object category output by the classification layer;

[0100] It should be noted that the object categories are target categories for model recognition and classification, which represent different types of objects or waste. In the solid waste recognition model, the object categories may include various types of solid waste, such as metal, plastic, glass, paper, organic waste, hazardous waste, etc., without limitation.

[0101] It should be further explained that the first prediction value refers to the preliminary prediction result given by the model based on the input data and the extracted features. In one embodiment, the first prediction value is a category probability or score, which indicates the model's confidence that each input sample belongs to each category.

[0102] Specifically, the element data features are input into the fully connected layer in the classification layer for classification to obtain a second prediction value, and then the second prediction value is subjected to category mapping to obtain a probability distribution value of each object category, thereby integrating each probability distribution value to generate the first prediction value.

[0103] Step S44: obtaining the solid waste identification model based on the first prediction value.

[0104] Specifically, the first prediction value, the preset sample label value and the preset weight matrix are calculated to obtain an error value, and then the error value is compared with a preset expected range. If the error value does not meet the preset expected range, the initial solid waste identification model is iteratively trained based on the error value to obtain the solid waste identification model.

[0105] In this embodiment, each of the second element data is input into the data vectorization layer to obtain several groups of third element data output by the data vectorization layer, and then the element data features in each of the third element data are extracted through the feature extraction layer, and then the element data features are input into the classification layer for classification to obtain the first prediction value of each object category output by the classification layer, and then based on the first prediction value, the solid waste recognition model is obtained, which helps the model to learn and recognize patterns more effectively, enables the model to process high-dimensional data, thereby improving classification accuracy and reducing the risk of model overfitting.

[0106] In a feasible implementation manner, inputting the element data feature into the classification layer for classification to obtain a first prediction value of each object category output by the classification layer includes:

[0107] Step S51, inputting the element data feature into the fully connected layer in the classification layer for classification to obtain a second prediction value;

[0108] It should be noted that the fully connected layer (FCL) refers to a layer in a neural network in which each neuron is connected to all neurons in the previous layer. In a convolutional neural network (CNN), the fully connected layer is usually located at the end of the network and is used to map the learned high-level features to the final output categories. In addition, in the fully connected layer, each input feature is considered independently, and the contribution of each input feature to each output node is weighted. The second prediction value refers to the original prediction result directly output by the fully connected layer.

[0109] Specifically, the element data feature is input into a classification layer, and the classification layer adopts 2 fully connected layers and adds a dropout method to reduce overfitting, so as to obtain a second prediction value.

[0110] Step S52, performing category mapping on the second prediction value to obtain a probability distribution value of each object category;

[0111] It should be noted that the probability distribution values ​​refer to normalized prediction values, which represent the model's estimate of the probability that the input data belongs to each category.

[0112] Specifically, the second prediction value is mapped to the category space, such as the output category is 33 categories, and then the output of the fully connected layer is normalized using the Softmax function to obtain the probability distribution value of each category, and the probability distribution value is directly used as the confidence, wherein the normalization is usually implemented through the softmax function, which converts the second prediction value (logits) into a probability value so that the sum of the probabilities of all categories is 1.

[0113] Step S53: Integrate the probability distribution values ​​to generate the first prediction value.

[0114] This embodiment obtains a second prediction value by inputting the element data feature into the fully connected layer in the classification layer for classification, and then classifies the second prediction value, and then classifies the second prediction value to obtain a probability distribution value of each object category, thereby integrating each probability distribution value to generate the first prediction value, and then learning the complex relationship between features through the fully connected layer to improve classification accuracy, ensure that the output values ​​are on the same scale, contribute to the stability and convergence of the model, and help decision makers understand the confidence of the model for each category, so as to make better decisions.

[0115] In a feasible implementation manner, obtaining the solid waste identification model based on the first prediction value includes:

[0116] Step S61, calculating the first prediction value, the preset sample label value and the preset weight matrix to obtain an error value;

[0117] It should be noted that the preset sample label value refers to the actual category or value corresponding to each sample when training and testing the machine learning model, so that the model's prediction results can be compared with the label value to evaluate its performance.

[0118] It should be further explained that the preset weight matrix refers to a matrix used to adjust the weights of samples of different categories in machine learning models, especially in classification problems. The purpose of this weight matrix is ​​to solve the problem of class imbalance or give more attention to certain categories according to business needs. The error value refers to the difference or gap between the model prediction value and the true label value.

[0119] Specifically, the first prediction value and the preset sample label value are calculated according to the error value calculation formula to obtain the error value. In one embodiment, in order to deal with the situation of sample imbalance, weights are introduced to balance the losses of different categories. The error value calculation formula is as follows:

[0120] H(p,q)=-∑ x ((γ l p(x)logq(x)))

[0121] Among them, H(p, q) is the cross entropy loss value between two probability distributions p and q, p and q represent the model prediction value and the actual label value respectively, γ l A fixed weight matrix will be calculated based on the actual data volume of different categories. When the sample size is large, γ l will be reduced, thus realizing the loss function feedback of imbalanced samples. p(x) is the true distribution (that is, the true probability of each category. Usually in classification problems, the probability of the correct category is 1 and the probability of other categories is 0), and q(x) is the predicted distribution, that is, the probability of each category predicted by the model.

[0122] Step S62, comparing the error value with a preset expected range;

[0123] It should be noted that the preset expected range refers to an acceptable range or threshold set for the error value during the model training process, which is used to determine whether the performance of the model has reached the expected standard. If the error value meets the preset expected range, the model is good enough and training can be stopped or other adjustments can be made; if the error value does not meet the preset expected range, the model needs further training or adjustment of model parameters.

[0124] Step S63: if the error value does not meet the preset expected range, then based on the error value, the initial solid waste identification model is iteratively trained to obtain the solid waste identification model.

[0125] It should be noted that if the error value does not meet the preset expected range, the initial solid waste identification model is iteratively trained based on the error value to obtain the solid waste identification model. In one embodiment, when the error value does not meet the preset expected range, the error value is transmitted back to the model, and the weight is updated according to the error value, and then through the continuous iterative training process, until the error value meets the preset expected range, the initial solid waste identification model whose error value meets the preset expected range is used as the solid waste identification model, so that the model gradually learns the patterns and laws in the data, and improves the prediction accuracy of solid waste elements. The weight update is carried out using the chain rule, that is, the partial derivative or deviation matrix of the weight is calculated, and the partial derivative value is added to the original weight to obtain a new weight matrix. In addition, the optimizer adopts an adaptive moment estimation method, which can estimate the step size of each parameter according to its historical gradient, so as to adaptively adjust the learning rate and update the model parameters.

[0126] In addition, after the solid waste recognition model is determined, the performance of the model can be evaluated through the validation set data, where the evaluation indicators include accuracy, recall rate, F1 score, etc.

[0127] This embodiment calculates the first predicted value, the preset sample label value and the preset weight matrix to obtain an error value, and then compares the error value with the preset expected range. If the error value does not meet the preset expected range, the initial solid waste identification model is iteratively trained based on the error value to obtain the solid waste identification model, and then dynamically adjusted according to the actual error value to adapt to the specific characteristics or changes of the data, thereby achieving continuous iterative training, gradually reducing the error between the predicted value and the true label, and thus improving the classification accuracy, while avoiding overfitting of the model on the training data, ensuring that the model has good generalization ability.

[0128] In a feasible implementation manner, after comparing the error value with a preset expected range, the method further includes:

[0129] Step S71, if the error value meets the preset expected range, the initial solid waste identification model is used as the solid waste identification model;

[0130] Specifically, after taking the initial solid waste identification model as the solid waste identification model, in order to improve the reliability of the evaluation, the solid waste identification model is evaluated by using a 10-fold cross-validation method, wherein cross-validation is a statistical analysis method that ensures the stability and accuracy of the evaluation results by dividing the data set into multiple parts and repeatedly training and evaluating the model on each part.

[0131] Step S72, obtaining the prediction accuracy and actual sample number corresponding to the solid waste identification model;

[0132] It should be noted that the prediction accuracy refers to the ratio of the number of samples correctly predicted by the model to the total number of samples. The actual number of samples refers to the number of samples actually trained on the model.

[0133] Step S73, based on the prediction accuracy and the actual number of samples, a confusion matrix diagram and a statistical box plot are generated to optimize the solid waste identification model.

[0134] It should be noted that the confusion matrix diagram is used to show the relationship between the model prediction results and the actual labels. It shows the correct and incorrect predictions of each category in the form of a matrix, where each cell of the confusion matrix diagram represents a combination of actual categories and predicted categories, such as true positives (TP), false positives (FP), true negatives (TN), and false negatives (FN), which helps to identify which categories the model performs well on, which categories need improvement, and whether the model has specific deviations.

[0135] It should be further explained that the statistical box plot is a statistical chart used to display the data distribution characteristics, so as to provide an intuitive display of the minimum value, first quartile (Q1), median (Q2), third quartile (Q3) and maximum value of the data, and can mark outliers in the data, thereby displaying the distribution of the model prediction values.

[0136] Specifically, the accuracy rate and the actual number of samples of the validation set are made into a confusion matrix diagram. In one embodiment, the X-axis of the confusion matrix diagram represents the actual category, the Y-axis represents the predicted category, and the diagonal line represents the correctly predicted category. The performance of the model is visualized by drawing the confusion matrix diagram, showing the prediction accuracy of the model for the waste to be identified.

[0137] In this embodiment, if the error value meets the preset expected range, the initial solid waste identification model is used as the solid waste identification model, and then the prediction accuracy and the actual number of samples corresponding to the solid waste identification model are obtained, and then a confusion matrix diagram and a statistical box diagram are generated based on the prediction accuracy and the actual number of samples to optimize the solid waste identification model, thereby ensuring that the performance of the model meets the expected standards, and intuitively displaying the performance of the model in each category through the confusion matrix diagram, including the number of correct classifications and incorrect classifications, helps to identify which categories the model performs poorly, thereby providing direction for model optimization. In addition, if it is found that the model performs poorly in certain categories, it can guide the collection of more data of related categories to improve the model.

[0138] For example, to help understand the implementation process of the solid waste identification method, please refer to Figure 4 , Figure 4 A simplified flow chart illustrating the solid waste identification method of this application.

[0139] It should be noted that the examples in the figures are only used to understand the present application and do not constitute a limitation on the solid waste identification method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0140] It should be understood that the order of execution of the steps in the above embodiment does not necessarily mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiment of the present invention.

[0141] This application also provides a solid waste identification device, please refer to Figure 5 , the solid waste identification device comprises:

[0142] An acquisition module 51 is used to acquire element detection data corresponding to the waste to be identified;

[0143] An output module 52, used for inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on several sets of conventional element data;

[0144] The classification module 53 is used to determine the target waste category of the waste to be identified based on the solid waste identification result.

[0145] The solid waste identification device is also used for:

[0146] Obtaining first element data of several groups of non-solid waste objects;

[0147] Performing random missing processing on each of the first element data to obtain several groups of second element data;

[0148] Each of the second element data is input into an initial solid waste identification model for iterative training to obtain the solid waste identification model.

[0149] The solid waste identification device is also used for:

[0150] Obtaining a preset missing element and an average value of missing elements corresponding to the preset missing element;

[0151] The first element data corresponding to the preset missing elements are randomly eliminated, and the missing element average values ​​are used for data filling to obtain several groups of second element data.

[0152] The solid waste identification device is also used for:

[0153] Input each of the second element data into the data vectorization layer to obtain a plurality of groups of third element data output by the data vectorization layer;

[0154] Extracting element data features in each of the third element data through the feature extraction layer;

[0155] Inputting the element data features into the classification layer for classification, and obtaining a first prediction value of each object category output by the classification layer;

[0156] Based on the first prediction value, the solid waste identification model is obtained.

[0157] The solid waste identification device is also used for:

[0158] Inputting the element data feature into the fully connected layer in the classification layer for classification to obtain a second prediction value;

[0159] Performing category mapping on the second predicted value to obtain a probability distribution value of each object category;

[0160] The probability distribution values ​​are integrated to generate the first prediction value.

[0161] The solid waste identification device is also used for:

[0162] Calculating the first prediction value, the preset sample label value, and the preset weight matrix to obtain an error value;

[0163] Comparing the error value with a preset expected range;

[0164] If the error value does not meet the preset expected range, the initial solid waste identification model is iteratively trained based on the error value to obtain the solid waste identification model.

[0165] The solid waste identification device is also used for:

[0166] If the error value meets the preset expected range, the initial solid waste identification model is used as the solid waste identification model;

[0167] Obtaining the prediction accuracy and actual sample number corresponding to the solid waste identification model;

[0168] Based on the prediction accuracy and the actual number of samples, a confusion matrix diagram and a statistical box plot are generated to optimize the solid waste identification model.

[0169] The solid waste identification device provided by the present application adopts the solid waste identification method in the above embodiment, and can solve the technical problems in the background technology. Compared with the prior art, the beneficial effects of the solid waste identification device provided by the present application are the same as the beneficial effects of the solid waste identification method provided by the above embodiment, and the other technical features in the solid waste identification device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0170] The present application provides a solid waste identification device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the solid waste identification method in the above-mentioned embodiment one.

[0171] Reference below Figure 6 , which shows a schematic diagram of the structure of a solid waste identification device suitable for implementing the embodiment of the present application. The solid waste identification device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 6 The solid waste identification device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0172] like Figure 6As shown, the solid waste identification device may include a processing device 1001 (such as a central processing unit, a graphics processor, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the solid waste identification device are also stored. The processing device 1001, ROM1002 and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the solid waste identification device to communicate wirelessly or wired with other devices to exchange data. Although the solid waste identification device with various systems is shown in the figure, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or have alternatively.

[0173] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0174] The solid waste identification device provided by the present application adopts the solid waste identification method in the above embodiment, which can solve the technical problems in the background technology. Compared with the prior art, the beneficial effects of the solid waste identification device provided by the present application are the same as the beneficial effects of the solid waste identification method provided by the above embodiment, and the other technical features in the solid waste identification device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0175] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0176] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0177] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the solid waste identification method in the above-mentioned embodiment.

[0178] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0179] The computer-readable storage medium may be included in the solid waste identification device; or may exist independently without being assembled into the solid waste identification device.

[0180] The computer-readable storage medium carries one or more programs. When the one or more programs are executed by the solid waste identification device, the solid waste identification device:

[0181] Obtaining element detection data corresponding to the waste to be identified;

[0182] Inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on a plurality of groups of first element data of non-solid waste objects;

[0183] Based on the solid waste identification result, a target waste category of the waste to be identified is determined.

[0184] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0185] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0186] The modules involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the module does not constitute a limitation on the unit itself in some cases.

[0187] The readable storage medium provided in the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned solid waste identification method, and can solve the technical problems in the background technology. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in the present application are the same as the beneficial effects of the solid waste identification method provided in the above-mentioned embodiment, and will not be repeated here.

[0188] An embodiment of the present application provides a computer program product, including a computer program, which implements the steps of the solid waste identification method as described above when executed by a processor.

[0189] The computer program product provided in this application can solve the technical problems in the background technology. Compared with the prior art, the beneficial effects of the computer program product provided in the embodiment of this application are the same as the beneficial effects of the solid waste identification method provided in the above embodiment, which will not be repeated here.

[0190] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A solid waste identification method, characterized in that: include: Obtaining element detection data corresponding to the waste to be identified; Inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on a plurality of groups of first element data of non-solid waste objects; Based on the solid waste identification result, a target waste category of the waste to be identified is determined.

2. The solid waste identification method according to claim 1, characterized in that: The training process of the solid waste identification model includes: Obtaining first element data of several groups of non-solid waste objects; Performing random missing processing on each of the first element data to obtain several groups of second element data; Each of the second element data is input into an initial solid waste identification model for iterative training to obtain the solid waste identification model.

3. The solid waste identification method according to claim 2, characterized in that: The randomly missing processing is performed on each of the first element data to obtain a plurality of groups of second element data, including: Obtaining a preset missing element and an average value of missing elements corresponding to the preset missing element; The first element data corresponding to the preset missing elements are randomly eliminated, and the missing element average values ​​are used for data filling to obtain several groups of second element data.

4. The solid waste identification method according to claim 2, characterized in that: The initial solid waste identification model includes a data vectorization layer, a feature extraction layer and a classification layer; The step of inputting each of the second element data into an initial solid waste identification model for iterative training to obtain the solid waste identification model comprises: Input each of the second element data into the data vectorization layer to obtain a plurality of groups of third element data output by the data vectorization layer; Extracting element data features in each of the third element data through the feature extraction layer; Inputting the element data features into the classification layer for classification, and obtaining a first prediction value of each object category output by the classification layer; Based on the first prediction value, the solid waste identification model is obtained.

5. The solid waste identification method according to claim 4, characterized in that: The step of inputting the element data feature into the classification layer for classification to obtain a first prediction value of each object category output by the classification layer includes: Inputting the element data feature into the fully connected layer in the classification layer for classification to obtain a second prediction value; Performing category mapping on the second predicted value to obtain a probability distribution value of each object category; The probability distribution values ​​are integrated to generate the first prediction value.

6. The solid waste identification method according to claim 4, characterized in that: The step of obtaining the solid waste identification model based on the first prediction value includes: Calculating the first prediction value, the preset sample label value, and the preset weight matrix to obtain an error value; Comparing the error value with a preset expected range; If the error value does not meet the preset expected range, the initial solid waste identification model is iteratively trained based on the error value to obtain the solid waste identification model.

7. The solid waste identification method according to claim 6, characterized in that: After comparing the error value with the preset expected range, the method further includes: If the error value meets the preset expected range, the initial solid waste identification model is used as the solid waste identification model; Obtaining the prediction accuracy and actual sample number corresponding to the solid waste identification model; Based on the prediction accuracy and the actual number of samples, a confusion matrix diagram and a statistical box plot are generated to optimize the solid waste identification model.

8. A solid waste identification device, characterized in that: include: An acquisition module, used for acquiring element detection data corresponding to the waste to be identified; An output module, used for inputting the element detection data into a solid waste identification model to obtain a solid waste identification result output by the solid waste identification model, wherein the solid waste identification model is obtained by iterative training based on several sets of conventional element data; A classification module is used to determine the target waste category of the waste to be identified based on the solid waste identification result.

9. A solid waste identification device, characterized in that: The solid waste identification device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the solid waste identification method according to any one of claims 1 to 7.

10. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the solid waste identification method according to any one of claims 1 to 7 are implemented.

Images