Intelligent sorting method for hazardous waste based on machine vision

Abstract

The application relates to the technical field of image processing, and discloses a hazardous waste intelligent sorting method based on machine vision, which comprises the following steps: obtaining a hazardous waste image to be sorted and extracting a multi-scale feature map; calculating an asymmetric weight factor, generating a gradient adjustment coefficient, weighting four distances by using the asymmetric weight factor and the gradient adjustment coefficient, and obtaining a center degree score of a pixel point; fusing the center degree score and a classification confidence to obtain a comprehensive sorting index; determining the number K of high-score pixel points, selecting K high-score pixel points by using an iterative screening strategy according to the comprehensive sorting index; performing category classification, boundary box regression and contour frequency domain regression based on the features of the selected high-score pixel point set; and jointly outputting the category, boundary box position and refined contour of the hazardous waste. The application not only improves the regression accuracy of the boundary box, but also realizes the extraction of the refined contour in a light-weighted manner, thereby solving the problem that the calculation amount is large and real-time application is difficult.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, in particular to a dangerous waste intelligent sorting method based on machine vision. BACKGROUND

[0002] Automatic and intelligent sorting is a link for improving processing efficiency and reducing manual operation risk in the field of hazardous waste (referred to as hazardous waste) treatment. The hazardous waste target detection technology based on computer vision provides an effective solution. The hazardous waste target detection technology based on computer vision can automatically identify the category and position of hazardous waste products on the conveying belt.

[0003] The hazardous waste target detection algorithm based on computer vision in the prior art is mainly divided into two stages and a single stage. The two-stage algorithm such as FasterR-CNN algorithm has the core idea that the first stage generates candidate regions, and the second stage classifies and regresses the boundary box of the candidate regions. In the first stage, a region candidate network is used to quickly generate high-quality candidate regions. RPN slides on the feature map, predicts which of the multiple anchor boxes is a foreground target for each position, and performs preliminary regression on the boundary box of the foreground target. In the second stage, the candidate regions generated by the RPN are sent to the RoI Pooling layer, and the candidate region features of different sizes are cropped and aligned to a fixed size. Then, these fixed-size feature maps are sent to the subsequent fully connected layer for final target classification and accurate boundary box regression. The detection accuracy of the FasterR-CNN algorithm is high, but the speed is slow, which is difficult to meet the real-time requirements of the flow line.

[0004] The single-stage algorithm such as YOLO, SSD, and FCOS has excellent detection speed in industrial scenarios and is suitable for industrial real-time applications. The FCOS algorithm is a single-stage and anchor-free target detection algorithm. It discards the traditional anchor mechanism and directly predicts the distance from each pixel point on the feature map to the four edges of the target boundary box. However, the FCOS algorithm has the problem of low boundary box detection accuracy, which leads to low accuracy of the detection result. SUMMARY

[0005] The present application provides a hazardous waste intelligent sorting method based on machine vision to solve the problems of slow detection speed in the prior art, which is difficult to meet the real-time requirements of the flow line, and low boundary box detection accuracy, which leads to low accuracy of the detection result.

[0006] The hazardous waste intelligent sorting method based on machine vision of the present application comprises the following steps:

[0007] The image of the hazardous waste to be sorted is obtained, and multi-scale feature maps are extracted using a backbone network and a feature pyramid network. For any pixel in the multi-scale feature map, an asymmetric weighting factor is calculated based on the distance predicted by the pixel to the four sides of the bounding box. A gradient adjustment coefficient is generated by combining the gradient information of the local neighborhood of the pixel. The four distances are weighted using the asymmetric weighting factor and the gradient adjustment coefficient to obtain the centrality score of the pixel.

[0008] The centrality score and classification confidence score are fused to obtain a comprehensive ranking index. The number of high-scoring pixels K is determined based on the size and category prior information of hazardous waste. Based on the comprehensive ranking index, an iterative screening strategy is used to select K high-scoring pixels that are spatially distributed. The iterative screening strategy is as follows: after selecting a highest-scoring pixel each time, Gaussian decay is applied to the comprehensive ranking index of other candidate pixels in the spatial neighborhood of the pixel.

[0009] Based on the features of the selected set of high-resolution pixels, category classification, bounding box regression, and contour frequency domain regression are performed. The contour frequency domain regression branch regresses the discrete Fourier transform coefficients of the hazardous waste contour from the features and reconstructs the refined contour through inverse transformation. The combined output includes the hazardous waste category, bounding box position, and refined contour.

[0010] Preferably, the step of weighting the four distances using the asymmetric weighting factor and gradient adjustment coefficient to obtain the centrality score of the pixel includes: calculating the distances predicted by the pixel to the left, right, top, and bottom edges of the bounding box. ; Calculate the horizontal distance ratio and vertical distance ratio ; for the feature map where the pixel is located The gradient magnitude is calculated using the Sober operator in the neighborhood; the centrality score of a pixel is obtained by the formula. The calculation yields C, where C is the centrality score. The ratio of horizontal distances. G represents the vertical distance ratio, and G is the gradient magnitude.

[0011] Preferably, the step of fusing the centrality score and classification confidence score to obtain a comprehensive ranking index includes: obtaining the centrality score and classification confidence score corresponding to the pixel; and using the geometric mean of the centrality score and classification confidence score as the comprehensive ranking index of the pixel.

[0012] Preferably, determining the number K of high-resolution pixels based on prior information about the size and category of hazardous waste includes: for each detected hazardous waste, obtaining the predicted category and predicted bounding box area of ​​the hazardous waste; based on the predicted category, searching for the corresponding basic quantity and size coefficient from a preset category configuration; and determining the number of high-resolution pixels. wherein D represents a predicted category, A represents a predicted bounding box area, represents a basic number corresponding to a predicted category, represents a size coefficient corresponding to a predicted category, K is rounded, and the value of K is limited in a preset range corresponding to the predicted category.

[0013] Preferably, the K high-score pixel points with spatial distribution are selected by using an iterative screening strategy, including: arranging all candidate pixel points in descending order according to the comprehensive ranking index; iterating K times, and in each iteration, selecting the pixel point with the highest ranking and adding the pixel point to the high-score pixel point set, and applying Gaussian attenuation to the comprehensive ranking index of all remaining candidate pixel points.

[0014] Preferably, the contour frequency domain regression branch regresses the discrete Fourier transform coefficients of the hazardous waste contour from the features, including: fusing the features of the K high-score pixel points into a fused feature vector through an average pooling operation; inputting the fused feature vector into a regression head network, wherein the regression head network outputs a 128-dimensional vector, and the vector represents the real part and the imaginary part of 64 Fourier coefficients of the hazardous waste contour.

[0015] Preferably, the fine contour is reconstructed by inverse transformation, including: reorganizing the 128-dimensional vector into 64 complex Fourier coefficients; applying inverse discrete Fourier transform to calculate 64 sampling point coordinates on the contour from the 64 complex Fourier coefficients; and connecting the 64 sampling point coordinates in order to form a closed fine contour.

[0016] Preferably, the backbone network is a ResNet-50 network.

[0017] Preferably, the Gaussian attenuation is calculated based on the Euclidean distance between the remaining candidate pixel points and the newly selected high-score pixel point.

[0018] Preferably, the standard deviation of the Gaussian function of the Gaussian attenuation is proportional to the size of the hazardous waste predicted bounding box.

[0019] ​The beneficial effects of the present application are: the present application combines the gradient adjustment coefficient of the pixel point and the asymmetric weight factor to calculate the center degree score, so that the center degree score can better reflect the positioning quality of the prediction box, and the regression accuracy of the bounding box is significantly improved for irregularly shaped hazardous waste. The center degree score and the classification confidence are fused into a comprehensive ranking index, and the high comprehensive ranking index pixel points are selected as high score pixel points according to the iterative screening strategy based on the comprehensive ranking index, so that a high score pixel point set with reasonable spatial distribution and low information redundancy can be obtained, and the training efficiency and detection performance of the network are improved. In addition, the contour frequency domain regression branch in the present application realizes the extraction of fine contours in a lightweight manner by predicting the Fourier coefficients of the contour and inversely transforming and reconstructing, thereby solving the problem of large calculation amount and difficult real-time application. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 A flowchart of a hazardous waste intelligent sorting method based on machine vision provided by an embodiment of the present application is shown. DETAILED DESCRIPTION

[0021] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present application, and cannot be understood as a limitation of the present application.

[0022] As shown in Figure 1 , the hazardous waste intelligent sorting method based on machine vision provided by the embodiment of the present application specifically includes the following steps:

[0023] S1, obtaining a hazardous waste image to be sorted, and extracting multi-scale feature maps by using a backbone network and a feature pyramid network.

[0024] Specifically, a hazardous waste image on a conveying belt is collected, and the hazardous waste image is input into a detection model with a ResNet-50 network as a backbone; after the ResNet-50 network outputs feature maps of different levels, a top-down and horizontal connection operation is performed by a feature pyramid network to fuse strong semantic information of a high layer and rich spatial details of a bottom layer, and five multi-scale feature maps of P3 to P7 are generated, which are respectively used for detecting hazardous waste of different sizes.

[0025] S2, for any pixel point on the multi-scale feature map, an asymmetric weight factor is calculated based on the distance of the pixel point prediction to the four sides of the bounding box, a gradient adjustment coefficient is generated by combining the gradient information of the local neighborhood of the pixel point, and the four distances are weighted by using the asymmetric weight factor and the gradient adjustment coefficient to obtain the center degree score of the pixel point.

[0026] S3, fuse the centrality score and the classification confidence to obtain a comprehensive ranking index; determine the number K of high-score pixel points according to the size and category prior information of the hazardous waste, and select K high-score pixel points with spatial distribution according to the comprehensive ranking index by using an iterative screening strategy; the iterative screening strategy is that after selecting a highest-score pixel point each time, a Gaussian attenuation is applied to the comprehensive ranking indexes of other candidate pixel points in the spatial neighborhood of the pixel point.

[0027] Specifically, at each iteration, the pixel point with the highest comprehensive ranking index is selected as the first high-score pixel point, and a Gaussian function value is multiplied to the comprehensive ranking indexes of other candidate pixel points within a certain spatial radius range with the pixel point as the center, so that the Gaussian function value increases with the increase of distance, thereby reducing the probability of selecting the neighboring points of the high-score pixel point in the next iteration. Repeat this process to select K high-score pixel points with uniform spatial distribution.

[0028] S4, based on the features of the selected high-score pixel point set, perform category classification, bounding box regression and contour frequency domain regression; wherein the contour frequency domain regression branch regresses the discrete Fourier transform coefficients of the hazardous waste contour from the features, and reconstructs the refined contour through inverse transform; and the category, bounding box position and refined contour of the hazardous waste are jointly output.

[0029] Specifically, the features of the selected K high-score pixel points are respectively input into three parallel prediction heads, and each prediction head is composed of four convolutional layers. The three parallel prediction heads are divided into a category classification head, a bounding box regression head and a contour frequency domain regression head. The category classification head is used to output the category of the hazardous waste, such as waste batteries, waste light tubes, etc. The bounding box regression head is used to output the center point coordinates and height and width of the target box. The contour frequency domain regression head is used to predict the real part and imaginary part of the first M low-frequency coefficients of the discrete Fourier transform DFT of the target contour, and the N contour vertex coordinates of the target can be efficiently reconstructed by using the predicted M Fourier coefficients through the fast inverse discrete Fourier transform IDFT algorithm, so as to realize the extraction of the refined contour.

[0030] In an optional embodiment, the four distances are weighted by using the asymmetric weight factor and the gradient adjustment coefficient to obtain the centrality score of the pixel point, including: calculating the distances of the pixel point prediction to the left, right, top and bottom edges of the bounding box; calculating the horizontal distance ratio and the vertical distance ratio ; the horizontal distance ratio and the vertical distance ratio constitute the asymmetric weight factor, and the The gradient magnitude G is calculated using the Sober operator in the neighborhood, and the gradient adjustment coefficient is calculated by adding 1 to the gradient magnitude G; the centrality score of a pixel is obtained by the formula... The calculation yields C, where C is the centrality score. The ratio of horizontal distances. G represents the vertical distance ratio, and G is the gradient magnitude.

[0031] Suppose that for a bounding box predicted as a battery, the size of the bounding box is... A pixel is the distance from a candidate pixel within the pixel to the left boundary. The distance to the right edge is 40 pixels. The distance from the top boundary is 160 pixels. The distance from the bottom boundary is 30 pixels. It is 70 pixels. Therefore, the horizontal distance is... =0.25, vertical distance ratio =0.43. These two ratios measure how centered a pixel is; the closer to 1, the more centered it is.

[0032] To determine whether a pixel is located within the hazardous waste outline region, the Sober operator is used to calculate the pixel's position on the feature map. Gradient magnitude in the neighborhood For example, if a pixel is located in the center of a vehicle's hood, the gradient magnitude will be higher because the color in the central area is smoother. Very low, for example, gradient magnitude The gradient magnitude is 0.1. Conversely, if the pixel happens to fall on the edge of the car window or the position of the headlight, the gradient magnitude is... It will be very high, for example, the gradient magnitude. The centrality score is 0.9. Substituting all factors into the formula to calculate the centrality score, the centrality score of the point on the hood is approximately 0.36; the centrality score of the point on the headlight is 0.62.

[0033] In an optional embodiment, fusing the centrality score and classification confidence score to obtain a comprehensive ranking index includes: obtaining the centrality score and classification confidence score corresponding to the pixel; and using the geometric mean of the centrality score and classification confidence score as the comprehensive ranking index of the pixel. .

[0034] In object detection tasks, a pixel has two metrics: a centrality score (C) and a classification confidence score (S). Using only one metric as the sole criterion may lead to suboptimal results. For example, a centrality score... For example, a pixel point with a classification confidence S of 0.95 is not a good representative pixel point because it is located in the blurred background, and the classification confidence S of the pixel point is only 0.2. For another example, a pixel point with a classification confidence S of 0.99 may be located at the edge of the target, and the center degree score C of the pixel point is only 0.1, so the pixel point is not suitable as a representative pixel point.

[0035] To balance the center degree score C and the classification confidence S, the embodiment fuses the center degree score C and the classification confidence S in a geometric mean manner. When the center degree score C and the classification confidence S of a pixel point are both high, the pixel point can obtain a higher comprehensive ranking index. Compared with an arithmetic mean, a geometric mean is more sensitive to a lower score, can effectively suppress the case of one high and one low, and screen out better candidate pixel points.

[0036] In an optional embodiment, the number K of high-score pixel points is determined according to prior information of the size and category of the hazardous waste, including: for each detected hazardous waste, obtaining a predicted category and a predicted bounding box area of the hazardous waste; according to the predicted category, searching for a corresponding basic number and a size coefficient from a preset category configuration; the number K of high-score pixel points is determined according to the following formula: wherein D represents the predicted category, A represents the predicted bounding box area, represents the basic number corresponding to the predicted category, represents the size coefficient corresponding to the predicted category, K is rounded, and the value of K is limited in a preset range corresponding to the predicted category .

[0037] The number of high-score pixel points required when describing different targets is different, so a fixed value cannot adapt to all targets. For example, describing a waste oil drum with a complex structure and a huge size requires more high-score pixel points than describing a battery with a simple structure and a small size. The embodiment establishes a category-related configuration table. For example, for a target with a category of waste oil drum, the basic number is 20, the corresponding size coefficient is 0.15, and the preset range is 15 to 50. For a target with a category of battery, the basic number is 10, the size coefficient is 0.1, and the preset range is 10 to 20.

[0038] For example, when a target is detected, it determines the target category and bounding box area. Suppose a waste oil drum is detected, and the bounding box area A of the waste oil drum is 2500 square pixels. According to the configuration table, the base quantity and size factor corresponding to the waste oil drum are looked up. Then, the bounding box area, base quantity, and size factor are substituted into the calculation formula for the number of high-resolution pixels K, and the value of K is equal to 27.5. After rounding, it is 28. The value is within the preset range of 15 to 50. Therefore, it is determined that 28 high-resolution pixels are selected.

[0039] In an optional embodiment, the step of selecting K high-resolution pixels with spatial distribution using an iterative screening strategy includes: sorting all candidate pixels in descending order according to a comprehensive ranking index; iterating K times, selecting the pixel with the highest current ranking in each iteration and adding it to the high-resolution pixel set; and applying Gaussian decay to the comprehensive ranking index of all other candidate pixels; the Gaussian decay is calculated based on the Euclidean distance between the other candidate pixels and the newly selected pixel, and the standard deviation of the Gaussian function is used. It is proportional to the size of the hazardous waste prediction bounding box.

[0040] To avoid the selected K high-resolution pixels all clustering in the same small area of ​​the target, and thus to obtain a spatially uniform set of high-resolution pixels that can represent the entire target shape, this embodiment employs an iterative suppression screening process. All candidate pixels located within the target bounding box are sorted from highest to lowest according to their comprehensive ranking index. In the first iteration, the top-ranked pixel P1 is unconditionally selected into the high-resolution pixel set. After selecting pixel P1, to prevent the next selected pixel from being too close to P1, the comprehensive ranking index of all other candidate pixels is updated. For any candidate pixel P... i Calculate the Euclidean distance between it and P1 on the image. P i The comprehensive ranking index after Gaussian decay is P i The initial overall ranking index is calculated by multiplying it by a Gaussian decay factor. The standard deviation σ of the Gaussian function is proportional to the size of the target bounding box. For example, for a pixel 5 pixels away from P1, the overall ranking index might be decayed to 30% of its original value, while for a pixel 50 pixels away from P1, the overall ranking index might only be decayed to 95% of its original value. After updating all overall ranking indices, the pixel P2 with the highest current overall ranking index is selected in the second iteration. P2 must be a pixel with a relatively high overall ranking index and a certain distance from P1. This process is repeated. Each time, the current best pixel is selected, and the neighboring pixels of the best pixel are suppressed.

[0041] In an optional embodiment, the contour frequency domain regression branch regresses discrete Fourier transform coefficients of the hazardous waste contour from the features, including: fusing the features of the K high-scored pixels into a fused feature vector through average pooling operation; inputting the fused feature vector into a regression head network, the regression head network outputting a 128-dimensional vector, the vector representing real parts and imaginary parts of 64 Fourier coefficients of the hazardous waste contour.

[0042] After selecting K representative high-scored pixels through the iterative screening strategy, the visual information contained in these high-scored pixels is used to describe the accurate contour of the target. Each high-scored pixel has a corresponding feature vector in the feature map of the deep neural network, such as a 256-dimensional vector, which encodes the local image pattern around the high-scored pixel. This embodiment performs average pooling on the 256-dimensional feature vectors of the K high-scored pixels, adds the first dimension of all K vectors and divides by K, adds the second dimension and divides by K, and so on, to generate a 256-dimensional fused feature vector.

[0043] The fused feature vector is sent to the regression head network, which is composed of several fully connected layers and can map the input 256-dimensional fused feature vector to a 128-dimensional output vector. The 128-dimensional vector is a feature representation of the target contour in the frequency domain. Specifically, the first 64 elements of the 128-dimensional vector are interpreted as the real parts of 64 Fourier coefficients describing the contour shape, and the last 64 elements are interpreted as the imaginary parts of these coefficients.

[0044] In an optional embodiment, the reconstruction of the refined contour through inverse transformation includes: reorganizing the 128-dimensional vector into 64 complex Fourier coefficients; applying inverse discrete Fourier transform to calculate 64 sample point coordinates on the contour from the 64 complex Fourier coefficients; and connecting the 64 sample point coordinates in order to form a closed refined contour.

[0045] Specifically, the 0th element and the 64th element of the vector are combined into a first complex Fourier coefficient, where the former is the real part and the latter is the imaginary part. Similarly, the 1st element and the 65th element form a second complex Fourier coefficient, and so on until all 64 complex Fourier coefficients are generated. This series of complex Fourier coefficients completely describes the contour shape of the target in the frequency domain. Applying an inverse discrete Fourier transform to the 64 complex Fourier coefficients can convert the frequency domain signal back to the time domain or spatial domain signal. In this embodiment, the output of the inverse discrete Fourier transform is a set of 64 complex numbers, and the real and imaginary parts of these complex numbers are interpreted as the x and y coordinates of the 64 sampling points on the contour, respectively. For example, the inverse discrete Fourier transform may produce the sequence (125.3+89.1i), (128.0+92.4i),..., which corresponds to the contour points (125.3, 89.1), (128.0, 92.4), etc. Connecting the 64 coordinate points in order according to their sequence in the sequence with straight lines, and connecting the last point with the first point, forms a closed polygon. This polygon is the refined contour of the target, which can more accurately depict the true shape of the target than the bounding box.

[0046] The implementation principle of the hazardous waste intelligent sorting method based on machine vision in the embodiment of the present application is that: the center degree score is calculated by combining the gradient adjustment coefficient of the pixel point and the asymmetric weight factor, so that the center degree score can better reflect the positioning quality of the prediction box, and the regression accuracy of the bounding box is significantly improved. In addition, the center degree score and the classification confidence are fused into a comprehensive ranking index, and the high comprehensive ranking index pixel points are selected as high score pixel points according to the comprehensive ranking index using an iterative screening strategy, so that a high score pixel point set with reasonable spatial distribution and low information redundancy can be obtained, thereby improving the training efficiency and detection performance of the network. Moreover, the contour frequency domain regression branch in the present application realizes the extraction of the refined contour in a lightweight manner by predicting the Fourier coefficients of the contour and inverse transforming and reconstructing, thereby solving the problem of large calculation amount and difficulty in real-time application.

[0047] Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A machine vision-based intelligent sorting method for hazardous waste, characterized in that, Includes the following steps: The system acquires images of hazardous waste to be sorted and extracts multi-scale feature maps using a backbone network and a feature pyramid network. For any pixel in the multi-scale feature map, an asymmetric weight factor is calculated based on the distance predicted by the pixel to the four sides of the bounding box, and a gradient adjustment coefficient is generated by combining the gradient information of the local neighborhood of the pixel. The centrality score of a pixel is obtained by weighting the four distances using an asymmetric weighting factor and a gradient adjustment coefficient. This includes calculating the predicted distances of the pixel to the left, right, top, and bottom edges of the bounding box. ; Calculate the horizontal distance ratio and vertical distance ratio ; for the feature map where the pixel is located The gradient magnitude is calculated using the Sober operator in the neighborhood; the centrality score of a pixel is obtained by the formula. The calculated value is C, which represents the centrality score. The ratio of horizontal distances. G represents the vertical distance ratio, and G is the gradient magnitude. The centrality score and classification confidence score are combined to obtain the comprehensive ranking index. The number of high-scoring pixels K is determined based on the size and category prior information of hazardous waste. Based on the comprehensive ranking index, an iterative screening strategy is adopted to select K high-scoring pixels that are spatially distributed. The iterative screening strategy is as follows: after selecting a highest-scoring pixel each time, Gaussian decay is applied to the comprehensive ranking index of other candidate pixels in the spatial neighborhood of the pixel. Based on the features of the selected set of high-resolution pixels, category classification, bounding box regression, and contour frequency domain regression are performed. The contour frequency domain regression branch regresses the discrete Fourier transform coefficients of the hazardous waste contour from the features and reconstructs the refined contour through inverse transformation. The combined output includes the hazardous waste category, bounding box position, and refined contour.

2. The intelligent hazardous waste sorting method based on machine vision according to claim 1, characterized in that, The method of fusing centrality score and classification confidence score to obtain a comprehensive ranking index includes: Obtain the centrality score and classification confidence score corresponding to the pixel; The geometric mean of the centrality score and the classification confidence score is used as the comprehensive ranking index for the pixels.

3. The intelligent hazardous waste sorting method based on machine vision according to claim 1, characterized in that, The determination of the number K of high-resolution pixels based on prior information about the size and category of hazardous waste includes: For each detected hazardous waste, obtain the predicted category and predicted bounding box area of ​​the hazardous waste; Based on the predicted category, the corresponding basic quantity and size factor are retrieved from the preset category configuration; Number of high-resolution pixels Where D represents the prediction category and A represents the predicted bounding box area. This represents the base number corresponding to the predicted category. The size factor corresponding to the predicted category is K, which is rounded down and its value is limited to a preset range corresponding to the predicted category. Inside.

4. The intelligent hazardous waste sorting method based on machine vision according to claim 1, characterized in that, The step of selecting K high-resolution pixels with spatial distribution using an iterative screening strategy includes: Sort all candidate pixels in descending order based on a comprehensive ranking metric; The algorithm iterates K times. In each iteration, the pixel with the highest current ranking is selected and added to the set of high-scoring pixels. Gaussian decay is applied to the overall ranking index of all other candidate pixels.

5. The intelligent hazardous waste sorting method based on machine vision according to claim 1, characterized in that, The contour frequency domain regression branch regresses the discrete Fourier transform coefficients of the hazardous waste contour from the features, including: The features of K high-resolution pixels are fused into a fused feature vector through average pooling. The fused feature vector is input into a regression head network, which outputs a 128-dimensional vector representing the real and imaginary parts of the 64 Fourier coefficients of the hazardous waste profile.

6. The intelligent hazardous waste sorting method based on machine vision according to claim 5, characterized in that, The process of reconstructing a refined contour through inverse transformation includes: The 128-dimensional vector is reorganized into 64 complex Fourier coefficients; The coordinates of 64 sampling points on the contour are calculated from the 64 complex Fourier coefficients by applying the inverse discrete Fourier transform. The coordinates of the 64 sampling points are connected sequentially to form a closed, refined profile.

7. The intelligent hazardous waste sorting method based on machine vision according to claim 1, characterized in that, The backbone network is a ResNet-50 network.

8. The intelligent hazardous waste sorting method based on machine vision according to claim 4, characterized in that, The Gaussian attenuation is calculated based on the Euclidean distance between the remaining candidate pixels and the newly selected high-resolution pixel.

9. The intelligent hazardous waste sorting method based on machine vision according to claim 8, characterized in that, The standard deviation of the Gaussian function of the Gaussian decay It is proportional to the size of the hazardous waste prediction bounding box.

Images