Classification device, selection device and classification method

Abstract

To provide a classification device capable of classifying plastic wastes according to a standard other than weight, specific gravity or size.SOLUTION: A classification device classifies plastic wastes. The classification device comprises a reception part and a processing device. The reception part receives an image obtained by photographing plastic wastes in a photographic region. The processing device extracts individual images by the plastic wastes from the image that the reception part receives, and classifies the plastic wastes according to a result of analysis of an object of analysis by the use of an analytic model structured by machine learning or a rule-based analytic program, the object of analysis being a surface shape or the whole shape of the plastic wastes appearing in the extracted individual images.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present application primarily relates to a sorting device for sorting plastic waste. [Background technology]

[0002] Patent Document 1 discloses a oscillating sorter for sorting industrial waste such as waste plastics. The oscillating sorter is equipped with a oscillating body. The industrial waste is placed on the oscillating body. As the oscillating body oscillates, a force is exerted on the industrial waste to move it forward. Furthermore, since the oscillating body is tilted upward at the front, a force is exerted on the industrial waste to move it backward due to its own weight. If the industrial waste is light, the force due to its own weight is weaker, so the industrial waste moves forward and falls from the light waste discharge port. On the other hand, if the industrial waste is heavy, the force due to its own weight is stronger, so the industrial waste moves backward and falls from the heavy waste discharge port. Furthermore, holes are formed in the oscillating body, and small-diameter waste falls through the holes and is collected. In this way, the industrial waste is sorted into light, heavy, and small-diameter waste. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2017-213489 Summary of the Invention [Problem to be solved by the invention]

[0004] The sorting device in Patent Document 1 sorts industrial waste materials by weight, specific gravity, or size. However, when sorting plastic waste, there are cases where it is preferable to sort the waste materials by criteria other than weight, specific gravity, or size. However, the sorting device in Patent Document 1 cannot sort the plastic waste materials to meet this requirement.

[0005] The present application has been made in view of the above circumstances, and its main object is to provide a sorting device that can sort plastic waste based on criteria other than weight, specific gravity, or size. [Means for solving the problem]

[0006] The problem to be solved by the present application is as described above. Next, the means for solving this problem and the effects thereof will be explained.

[0007] According to a first aspect of the present application, there is provided a classification device having the following configuration. That is, the classification device classifies plastic waste. The classification device includes a receiving unit and a processing unit. The receiving unit receives images obtained by photographing the plastic waste in a photographing area. The processing unit extracts individual images for each piece of plastic waste from the images received by the receiving unit, analyzes the surface shape or overall shape of the plastic waste that appears in the extracted individual images, and classifies the plastic waste according to the results of analyzing the analysis target using an analytical model constructed by machine learning or a rule-based analytical program.

[0008] According to a second aspect of the present application, the following classification method is provided. That is, in the classification method, plastic waste is classified. In the classification method, the plastic waste in a photographing area is photographed to generate images. In the classification method, individual images for each of the plastic waste are extracted from the generated images. In the classification method, the surface shape or overall shape of the plastic waste that appears in the individual images is used as an analysis object, and the plastic waste is classified according to the results of analyzing the analysis object using an analytical model constructed by machine learning or a rule-based analytical program. [Effects of the Invention]

[0009] According to the present application, plastic waste can be sorted on criteria other than weight, specific gravity, or size. [Brief explanation of the drawings]

[0010] [Figure 1] 1 is a schematic diagram showing the configuration of a sorting device according to an embodiment of the present application. [Figure 2]1 is a flow chart showing a process for sorting plastic waste. [Figure 3] Table explaining the contents of material analysis and product analysis. DETAILED DESCRIPTION OF THE INVENTION

[0011] Next, an embodiment of the present application will be described with reference to the drawings.

[0012] The sorting device 1 shown in Figure 1 is a device that classifies and separates plastic waste, which is the target material, based on various criteria. "Classification" means distinguishing according to predetermined criteria. "Sorting" means moving plastic waste according to the classification results. Plastic waste is waste plastic products that contain plastic as a main component. Plastic waste may be supplied to the sorting device 1 after being compressed, crushed, or otherwise processed, or may be supplied to the sorting device 1 without being compressed, crushed, or otherwise processed. In either case, there is a high possibility that at least a partial shape of the plastic product remains in the plastic waste.

[0013] As shown in FIG. 1, the sorting device 1 includes a conveyor 10, a camera 12, a lighting device 13, a receiving unit 14, a processing device 15, and a sorting unit 16.

[0014] The conveyor 10 transports plastic waste supplied from the outside downstream. The conveyor 10 is, for example, a belt conveyor or a roller conveyor. Instead of the conveyor 10, other transport devices capable of transporting plastic waste may be used. The other transport devices may be, for example, arm robots that grab and move plastic waste. Note that transport devices such as the conveyor 10 are not essential components and may be omitted.

[0015] The camera 12 is disposed above the conveyor 10. The imaging area of ​​the camera 12 is the upper surface of the conveyor 10. The camera 12 captures an image of one or more plastic waste objects in the imaging area to generate an image. The camera 12 in this embodiment is a line camera. The camera 12 acquires line images of the plastic waste objects placed on the upper surface of the conveyor 10. The camera 12 or the processing device 15 generates an image showing one or more plastic waste objects based on the multiple line images. Note that the camera 12 is not limited to a line image camera, and may be a camera with a rectangular or circular imaging area. The image generated by the camera 12 may be an image in which a color is assigned to each pixel, or may be a distance image in which the distance from the imaging position is assigned to each pixel. When generating a distance image, it is preferable to use a stereo camera or a 3D sensor instead of a visible light camera.

[0016] The lighting device 13 is disposed above the conveyor 10. The lighting device 13 illuminates the shooting area of ​​the camera 12. This allows the darkened parts of the plastic waste to be clearly imaged. This makes it easier for the surface shape of the plastic waste to appear clearly in the image. Note that since images of the plastic waste can be generated without lighting, the lighting device 13 is not an essential component and can be omitted.

[0017] The receiving unit 14 is a communication module compatible with wired or wireless communication. The receiving unit 14 receives images generated by the camera 12. The images received by the receiving unit 14 are output to the processing device 15 for processing. The receiving unit 14 and the processing device 15 may be separate devices, or may be a single control device built into a single housing.

[0018] The processing device 15 includes an arithmetic device such as an FPGA, an ASIC, or a CPU, and a storage device such as an SSD or a flash memory. The processing device 15 reads out programs stored in the storage device and executes them with the arithmetic device, thereby performing various processes related to the sorting device 1. For example, the processing device 15 controls the conveyor 10 and the sorting unit 16, processes images generated by the camera 12, and analyzes these images. Details of the processes performed by the processing device 15 will be described later.

[0019] The sorting device 1 of this embodiment separates and separates plastic waste into soft plastics and hard plastics. When recycling plastic waste, it may be necessary to separate the plastic waste into soft plastics and hard plastics. This is because the companies that recycle soft plastics may differ from the companies that recycle hard plastics. Another reason is that the temperatures and heating times required to melt soft plastics and hard plastics differ, making melting difficult or inefficient when soft and hard plastics are mixed together.

[0020] Soft plastics are products that are very thin and flexible, such as plastic bags, films, and picnic sheets. Hard plastics are products that are relatively thick and inflexible, such as plastic toys, plastic cases, and hangers. However, classification standards for soft and hard plastics may differ depending on the recycler. Furthermore, recycling applications do not require strict standards or accurate classification.

[0021] The sorting unit 16 has a storage unit that stores compressed air and an injection unit that injects the compressed air. Multiple injection units are arranged in the width direction of the conveyor 10. The position where the injection unit is installed is called the sorting position. The sorting unit 16 injects compressed air at the plastic waste at the sorting position, blowing the plastic waste to the soft plastic recovery unit 17. On the other hand, if the sorting unit 16 does not inject compressed air, the plastic waste falls into the hard plastic recovery unit 18. In this way, the plastic waste can be sorted into soft plastics and hard plastics. Note that the sorting unit 16 is not an essential component and can be omitted. A device without the sorting unit 16 is used, for example, to detect the content ratio of soft plastics and hard plastics.

[0022] The sorting unit 16 may sort hard plastics by spraying compressed air at them. Furthermore, the sorting unit 16 is not limited to a machine that sprays compressed air. For example, the sorting unit 16 may be an arm robot. The arm robot picks up plastic waste and transports it to the soft plastic recovery unit 17 or the hard plastic recovery unit 18.

[0023] Next, details of the processing by the receiving unit 14 and the processing device 15 will be described with reference to Figures 2 and 3. The receiving unit 14 and the processing device 15 classify plastic waste based on images, and therefore correspond to the "classification device 20."

[0024] First, the receiving unit 14 receives an image of the shooting area from the camera 12 (S101). If the image received by the receiving unit 14 is a line image, the processing device 15 generates an image by combining a plurality of line images.

[0025] Next, the processing device 15 pre-adjusts the image of the photographed area (S102). Pre-adjustment is a process for making it easier to extract an image of plastic waste from the image of the photographed area. For example, plastic waste may contain black parts, which may be difficult to distinguish from shaded parts. Furthermore, if the conveyor surface is black, it may be difficult to identify the boundary between the black parts of the plastic waste and the conveyor surface. Therefore, the processing device 15 adjusts the parameters for image rendering so that the black parts of the plastic waste are emphasized. Furthermore, plastic waste may contain parts that strongly reflect light, and in order to reduce the influence of these parts, the processing device 15 may perform adjustments such as median processing. Note that pre-adjustment is not a required process and can be omitted.

[0026] Next, the processing device 15 extracts individual images of each piece of plastic waste from the image of the photographed area (S103). Specifically, the processing device 15 identifies the contours of the plastic waste through image processing. Because the conveyor surface and the plastic waste are basically different in color, individual images of the plastic waste can be extracted from the image of the photographed area based on the color difference. Note that if the colors of the conveyor surface and the plastic waste are similar, there is a possibility that an image of one piece of plastic waste will be extracted in parts. Therefore, if the pixel proximity of the extracted images is below a threshold, the processing device 15 may combine those images and treat them as an individual image of one piece of plastic waste.

[0027] Next, the processing device 15 selects one individual image from the multiple individual images extracted in step S103 (S104). The selected individual image becomes the analysis target. The processing device 15 analyzes all individual images, so the order of analysis is not particularly limited. Alternatively, plastic waste that is sorted early, in other words, plastic waste located downstream on the conveyor 10, may be analyzed early on. Furthermore, instead of analyzing individual images one by one, multiple individual images may be analyzed in parallel.

[0028] The processing device 15 then uses the individual images to perform material analysis and product analysis of the plastic waste.

[0029] Material analysis is a process of estimating the constituent materials of plastic waste. In this embodiment, for example, the material analysis estimates whether the plastic waste is soft plastic or hard plastic. Note that the classification of material analysis in this embodiment is an example.

[0030] The target of material analysis is the surface shape of plastic waste that appears in individual images. For example, soft plastics are flexible, so their surfaces are wavy or have irregular curves. On the other hand, hard plastics are not flexible, so their surfaces are flat or have relatively large flat surfaces.

[0031] The analytical method for material analysis is an analytical model or an analytical program. The analytical model is constructed by machine learning training data. The training data is, for example, data that associates images of plastic waste with information indicating whether the waste is soft or hard plastic. The machine learning is supervised learning. The analytical model constructed can be, for example, a type commonly used in image processing, such as logistic analysis or SVM. The output of the analytical model is a score for each constituent material that is an analysis candidate. This score is a numerical representation of the probability of belonging to each analysis candidate. Since the method of outputting the probability of belonging by using a model such as logistic analysis is well known, a detailed explanation will be omitted.

[0032] Another analytical method for material analysis is the use of an analysis program. The analysis program is constructed using a rule-based algorithm rather than AI. For example, when using an analysis program, the processing device 15 first identifies areas where surface waviness appears from individual images through image processing. Next, the processing device 15 calculates a score such that the greater the proportion of areas where waviness appears, the higher the probability that the plastic is soft, and the smaller the proportion of areas where waviness appears, the higher the probability that the plastic is hard. Alternatively, a score may be calculated based on the probability of the plastic being soft or hard, based on the proportion of areas where flat surfaces form within the plastic waste.

[0033] The product analysis estimates the type of product made from the plastic waste. The product analysis of this embodiment estimates whether the plastic waste falls into one of the following categories: plastic bags, films, hangers, plastic toys, etc. Note that the classification of the product analysis of this embodiment is an example, and it may also be possible to estimate whether the plastic waste falls into a container or a non-container category, for example.

[0034] The target of product analysis is the overall shape of plastic waste that appears in individual images. For example, the overall shapes of plastic bags, films, hangers, and plastic toys have distinctive shapes that can be distinguished from others.

[0035] The analytical method for product analysis is an analytical model or an analytical program. The same parts of the analytical method for product analysis as those for material analysis will not be explained here. The learning data for the analytical model for product analysis is, for example, data that associates images of plastic waste with information indicating the type of product shown in the image. The output of the analytical model is a score for each type of product that is a candidate for analysis.

[0036] Another analytical method for product analysis is the use of an analysis program. Before using the analysis program, a product database of plastic waste is created. The product database is a database that associates types of plastic products with corresponding images. It is preferable to register multiple images. When using the analysis program, the processing device 15 performs image matching between the individual images of plastic waste and the product database, and identifies images that are similar to the individual images of plastic waste. This allows the type of product of the plastic waste in the individual images to be inferred. The degree of match in the matching process corresponds to a score. A higher degree of match in the matching process indicates a higher probability of matching the product type.

[0037] Furthermore, the results of product analysis can be used to improve the accuracy of material analysis. In other words, there is a correlation between the type of product and the product's constituent material. For example, plastic bags are likely to be made of soft plastic, hangers are likely to be made of hard plastic, and plastic toys are likely to be made of hard plastic. Therefore, if the product analysis results indicate a plastic bag, the plastic waste in the individual image is likely to be soft plastic. Specifically, the results of product analysis are used in the following manner. In this case, the product database described above additionally registers whether the product's constituent material is soft or hard plastic in association with the product. By using the analysis program described above, it is possible to estimate whether the plastic waste in the individual image is soft or hard plastic based on the product analysis results and the information added to the product database. In this case, the value obtained by multiplying the score corresponding to the product analysis results by a predetermined multiplier is used as a correction value. Then, the final score, which is the sum of the score calculated by material analysis and the correction value, is used to estimate whether the plastic waste is soft or hard plastic. The predetermined multiplier is a value for adjusting the magnitude of the score calculated by the product analysis to the magnitude of the score calculated by the material analysis. Furthermore, the predetermined multiplier also has the meaning of weighting, which adjusts the extent to which the analysis results of the product analysis are reflected in the results.

[0038] In this embodiment, the results of the product analysis are used to assist the material analysis. That is, the processing device 15 calculates a score using the above-described material analysis (S105), calculates a correction value using the above-described product analysis (S106), and calculates a final score by adding the score and the correction value (S107). Next, the processing device 15 classifies the plastic waste into soft plastic or plastic waste depending on the magnitude of the final score (S108). This completes the processing for one individual image. Thereafter, the processing device 15 processes another individual image in the same way and classifies it as either soft plastic or hard plastic.

[0039] The processing device 15 stores the classification results in association with identification information for identifying the plastic waste. Thereafter, when the plastic waste reaches the sorting position, if the plastic waste is soft plastic, the sorting unit 16 is operated to blow it off to the soft plastic recovery unit 17, and if the plastic waste is hard plastic, the sorting unit 16 is not operated and the plastic waste falls into the hard plastic recovery unit 18.

[0040] As described above, plastic waste can be classified and separated into soft plastics and hard plastics. The method of Patent Document 1 does not allow for classification and separation according to the type of constituent material or product. A method of separating materials using wind power is also known, but even when wind power is used, separation is based on the weight or specific gravity of the objects, so classification and separation according to the type of constituent material or product is not possible.

[0041] In this embodiment, the score calculated in the material analysis is calculated using a correction value calculated in the product analysis. Specifically, the score calculated in the material analysis using the analytical model is corrected by the score calculated in the product analysis using the analytical program. This allows flexible analysis using machine learning to be utilized while being supplemented by rule-based analysis.

[0042] However, the correction value calculated in the product analysis is not required, and the material analysis may be performed using only the score calculated in the material analysis. In this case, the score may be calculated using an analytical model or an analytical program. Also, the sorting device 1 may be used to classify and sort products according to the type of product rather than the constituent materials. In this case, the score may be calculated using an analytical model or an analytical program.

[0043] (Feature 1) As described above, the classification device 20 of this embodiment classifies plastic waste. The classification device 20 includes a receiving unit 14 and a processing device 15. The receiving unit 14 receives images obtained by photographing plastic waste in an imaging area. The processing device 15 extracts individual images for each piece of plastic waste from the images received by the receiving unit 14, analyzes the surface shape or overall shape of the plastic waste that appears in the extracted individual images, and classifies the plastic waste according to the results of analyzing the analysis target using an analytical model constructed by machine learning or a rule-based analysis program.

[0044] This allows plastic waste to be classified according to the shape that appears in the individual images of the plastic waste, and therefore allows classification based on criteria other than weight, specific gravity, or size.

[0045] (Feature 2) In the classification device 20 of this embodiment, the processing device 15 classifies the plastic waste according to its constituent materials based on the results of analyzing the surface shape of the plastic waste appearing in the individual images using an analytical model or an analytical program.

[0046] It can meet the needs of processing plastic waste for each constituent material.

[0047] (Feature 3) In the sorting device 20 of this embodiment, the processing device 15 sorts plastic waste into items made of soft plastics and items made of hard plastics.

[0048] It can meet the needs of plastic waste disposal by separating it into soft and hard plastics.

[0049] (Feature 4) In the classification device 20 of this embodiment, the processing device 15 classifies the plastic waste according to the type of product based on the results of analyzing the overall shape of the plastic waste appearing in the individual images using an analytical model or an analytical program.

[0050] We can meet the needs of plastic waste disposal depending on the product and use.

[0051] (Feature 5) In the sorting device 20 of this embodiment, the processing device 15 sorts plastic waste into items whose product type corresponds to containers and other items.

[0052] It can meet the needs of treating plastic waste by separating it into containers and non-containers.

[0053] (Feature 6) In the classification device 20 of this embodiment, the processing device 15 classifies the plastic waste according to the results of analyzing the surface shape or overall shape of the plastic waste that appears in the individual images using an analytical model.

[0054] This allows the use of machine learning to classify plastic waste, taking into account aspects that cannot be determined by rule-based methods.

[0055] (Feature 7) In the classification device 20 of this embodiment, the processing device 15 classifies the plastic waste according to the results of analyzing the surface shape or overall shape of the plastic waste that appears in the individual images using an analysis program.

[0056] This allows plastic waste to be classified based on rules, allowing it to be classified according to pre-determined criteria.

[0057] (Feature 8) In the classification device 20 of this embodiment, the processing device 15 calculates a score corresponding to a classification candidate based on the results of analyzing the surface shape or overall shape of the plastic waste appearing in the individual images using an analytical model. The processing device 15 calculates a correction value corresponding to the classification candidate based on the results of analyzing the surface shape or overall shape of the plastic waste appearing in the individual images using an analysis program. The processing device 15 calculates a final score based on the sum of the score and the correction value. The processing device 15 classifies the plastic waste based on the final score.

[0058] Because correction values ​​are used from the analysis program, it is possible to accurately classify plastic waste that cannot be adequately classified using the analysis model alone.

[0059] (Feature 9) In the classification device 20 of this embodiment, the processing device 15 calculates a score that quantifies whether the plastic waste is likely to be soft plastic or hard plastic based on the results of analyzing the surface shape of the plastic waste appearing in the individual image using an analytical model. The processing device 15 calculates a correction value that quantifies whether the plastic waste is likely to be soft plastic or hard plastic based on the type of product obtained by analyzing the overall shape of the plastic waste appearing in the individual image using an analytical program. The processing device 15 calculates a final score based on the sum of the score and the correction value. The processing device 15 classifies the plastic waste as soft plastic or hard plastic based on the final score.

[0060] Because correction values ​​are used from the analysis program, plastic waste that cannot be adequately classified using the analysis model alone can be accurately classified as either soft plastic or hard plastic. (Feature 10) The sorting device 1 of this embodiment includes a sorting device 20, a camera 12, and a sorting unit 16. The camera 12 captures images. The sorting unit 16 moves the plastic waste to a collection position according to the sorting results obtained by the sorting device 20.

[0061] This allows plastic waste to be moved and collected at locations according to classification criteria other than weight, specific gravity, or size.

[0062] The above-described features 1 to 10 can be combined, for example, as follows to realize the classification device 20 or the selection device 1. The same applies to the classification method or the selection method. [Configuration 1] Classification device 20 for feature 1. [Configuration 2] A classification device 20 having feature 2 in addition to configuration 1. [Configuration 3] A classification device 20 having feature 3 in addition to configuration 2. [Configuration 4] A classification device 20 having any one of configurations 1 to 3, and further having feature 4. [Configuration 5] A classification device 20 having feature 5 in addition to configuration 4. [Configuration 6] A classification device 20 having any one of configurations 1 to 5, and further having feature 6. [Configuration 7] A classification device 20 having any one of configurations 1 to 6, and further having feature 7. [Configuration 8] A classification device 20 having feature 8 in addition to configuration 1. [Configuration 9] A classification device 20 having feature 9 in addition to configuration 1.

[0063] The preferred embodiment of the present application has been described above, but the above configuration can be modified, for example, as follows. Each modification may be made alone, or multiple modifications may be made in any combination.

[0064] The flowcharts shown in the above embodiments are merely examples, and some processes may be omitted, the contents of some processes may be changed, or new processes may be added.

[0065] In the above embodiment, plastic waste is classified and sorted into two types. Alternatively, plastic waste may be classified and sorted into three or more types. For example, plastic waste may be classified into container plastics, non-container and soft plastics, and non-container and hard plastics. In this case, both material analysis and product analysis are performed on the individual images.

[0066] The functions of the elements disclosed herein can be performed using circuits or processing circuitry, including general-purpose processors, special-purpose processors, integrated circuits, ASICs (Application Specific Integrated Circuits), conventional circuits, and / or combinations thereof, configured or programmed to perform the disclosed functions. A processor is considered a processing circuit or circuitry because it includes transistors and other circuitry. In this disclosure, a circuit, unit, or means is hardware that performs the recited functions or hardware that is programmed to perform the recited functions. The hardware may be hardware disclosed herein or other known hardware that is programmed or configured to perform the recited functions. Where the hardware is a processor, which is considered a type of circuit, the circuit, means, or unit is a combination of hardware and software, and the software is used to configure the hardware and / or processor. [Explanation of symbols]

[0067] 1. Sorting device 12 Camera 13 Lighting equipment 14 Receiving unit 15 Processing equipment 16 Sorting Department

Claims

1. A sorting device for sorting plastic waste, comprising: a receiving unit for receiving an image obtained by photographing the plastic waste in a photographing area; A processing device that extracts individual images for each of the plastic wastes from the images received by the receiving unit, analyzes the surface shape or overall shape of the plastic waste that appears in the extracted individual images, and classifies the plastic wastes according to the results of analyzing the analysis targets using an analytical model constructed by machine learning or a rule-based analytical program; A classification device comprising:

2. The classification device according to claim 1 , The processing device is a classification device that classifies the plastic waste according to its constituent materials based on the results of analyzing the surface shape of the plastic waste appearing in the individual images using the analytical model or the analytical program.

3. 3. The classification device according to claim 2, The processing device is a sorting device that sorts the plastic waste into items made of soft plastics and items made of hard plastics.

4. The classification device according to claim 1 , The processing device is a classification device that classifies the plastic waste according to the type of product based on the results of analyzing the overall shape of the plastic waste appearing in the individual images using the analytical model or the analytical program.

5. 5. The classification device according to claim 4, The processing device is a sorting device that sorts the plastic waste into products that correspond to containers and other products.

6. The classification device according to claim 1 , The processing device is a classification device that classifies the plastic waste according to the results of analyzing the surface shape or overall shape of the plastic waste that appears in the individual images using the analytical model.

7. The classification device according to claim 1 , The processing device is a classification device that classifies the plastic waste according to the results of analyzing the surface shape or overall shape of the plastic waste that appears in the individual images using the analysis program.

8. The classification device according to claim 1 , The processing device calculates a score according to a classification candidate according to a result of analyzing the surface shape or overall shape of the plastic waste appearing in the individual image using the analytical model, The processing device calculates a correction value according to the classification candidate based on the results of analyzing the surface shape or overall shape of the plastic waste appearing in the individual image using the analysis program, the processing device calculates a final score based on the sum of the score and the correction value; The processing device sorts the plastic waste based on the final score.

9. The classification device according to claim 1 , The processing device calculates a score that quantifies whether the surface shape of the plastic waste appearing in the individual image is likely to be soft plastic or hard plastic according to the results of analyzing the surface shape of the plastic waste appearing in the individual image using the analytical model, The processing device calculates a correction value that quantifies whether the product is likely to be soft plastic or hard plastic based on the type of product obtained by analyzing the overall shape of the plastic waste appearing in the individual image using the analysis program, the processing device calculates a final score based on the sum of the score and the correction value; The processing device classifies the plastic waste into soft plastics or hard plastics based on the final score.

10. The classification device of claim 1; A camera for taking images; A sorting unit that moves the plastic waste to a collection position according to the classification result by the sorting device; A sorting device comprising:

11. A classification method for classifying plastic waste, comprising: Photographing the plastic waste in a photographing area to generate an image; Extracting individual images for each of the plastic wastes from the generated images; A classification method in which the surface shape or overall shape of the plastic waste that appears in the individual images is the object of analysis, and the plastic waste is classified according to the results of analyzing the object of analysis using an analytical model constructed by machine learning or a rule-based analytical program.

Images