KR102209147B1 - Automatic classification system of recycled products or used goods and automatic classification method using the same - Google Patents

Abstract

Disclosed are an automatic classification system for recycled products or used products and an automatic classification method using the same. The system interconnected through a network with a waste emitter terminal and a waste collector terminal of the present invention comprises: a communication device which receives a waste image transmitted from a waste emitter terminal, and emitter information related to waste, and receives collector information related to the waste transmitted from the waste collector terminal; and a classification device which classifies corresponding waste as recycled products or used products by using the waste image, the emitter information related to the waste, and the collector information related to the waste.

Description

Automatic classification system of recycled products or used goods and automatic classification method using the same}

The present invention relates to an automatic classification system for recyclables or used goods, and relates to an automatic classification method using the same, and more specifically, a recycled product that automatically classifies waste into used or recycled products using information on the waste secured from dischargers and collectors. Or it relates to an automatic classification system for used goods and an automatic classification method using the same.

Recognizing the importance of the environment, various approaches are being attempted in the way waste is treated. Among wastes other than designated wastes, the proportion of wastes treated by recycling is gradually increasing. However, depending on the type of waste, recycling is pointed out as a very difficult problem.

For example, only about 33% of waste households generated annually in Korea are recycled as plywood or wood materials, about 59% are incinerated, and the remaining 8% are neglected and dumped. Recycling waste furniture as a raw material for wood is positive in terms of resource recycling. However, if the waste furniture can be reused as its original function, it is more economical and environmentally advantageous than the resource recycling method.

However, in order to use waste such as waste furniture as a resource reuse method rather than a resource recycling method, the following technical difficulties exist. Waste should be classified as either a disposal method or a recycling method. Even if it is classified as a recycling treatment method, it is necessary to distinguish whether it is treated as a resource recycling method or a resource reuse method (use of used goods).

In order to dispose of the waste, the waste emitter pays the cost of treating the waste and requests the waste to be disposed of. Since it is difficult for waste dischargers to judge whether the waste is suitable for recycling or reuse, it is difficult to determine whether the waste is suitable for recycling or reuse, so it is a new type of recycled product that automatically classifies the waste as a recyclable or used product based on data obtained from the discharger and collector. There is a need to develop an automatic classification system.

Patent Document 1: Korean Patent Registration No. 10-1040177 (June 9, 2011) Patent Document 2: Korean Patent Registration No. 10-1783523 (September 29, 2017) Patent Document 3: Korean Patent Registration No. 10-1964459 (April 1, 2019) Patent Document 4: Korean Patent Publication No. 10-2019-0031435 (March 26, 2019)

The present invention, conceived by the above-described necessity, classifies the type of waste by using the waste image input when the waste discharger makes a request for waste treatment, and uses recycled products or used products based on information such as age of use and appearance of waste. The purpose of this is to provide an automatic classification system for recycled or used goods that automatically sorts them into goods and a method using the same.

In order to achieve the above object, as a system interconnected through a network with a waste discharger terminal and a waste collector terminal according to an embodiment of the present invention, the automatic sorting system for recycled or used goods includes waste images and wastes transmitted from the waste discharger terminal. A communication device for receiving information about a discharger and receiving information about a waste collector transmitted from the waste collector terminal; And a classification device for classifying and evaluating the waste as a recycled product or a used product by using the waste image, discharger information on the waste, and collector information on the waste.

In this case, the classification device analyzes the waste image or discharger information about the waste to classify the waste into furniture, household appliances, or other categories, and analyzes the discharger information about the waste Using the service life calculation unit that calculates the service life, the appearance condition identification unit that identifies the appearance state of the waste by analyzing the collector information about the waste, and the type of the waste, the service life and the information on the appearance state. It includes a processing method decision unit that classifies the waste as recycled or used goods.

In this case, the category classification unit compares the waste image information value included in the waste image with the reference image information value included in the plurality of machine-learned waste images to convert the waste into one of furniture, household appliances, or others. Classify.

On the other hand, the sorting apparatus classifies the type of waste into a matching category by matching a result value obtained by processing the discharger information about the waste in a natural language with a plurality of keyword values mapped to furniture, home appliances, or other categories.

Meanwhile, the service life calculation unit calculates service life information for the waste by using the purchase time information of the waste included in the discharger information and the recommended service life of the waste.

On the other hand, the appearance state identification unit calculates appearance state information for the waste by using the score assigned according to the appearance state of the waste included in the collector information and the user's feedback information.

Meanwhile, the treatment method determination unit calculates a classification result value obtained by multiplying the age information and the appearance state information by weights, respectively, and compares the calculated classification result value with a predetermined reference value to treat the waste as recycled or used goods. Judge the way.

An automatic classification method using a system interconnected through a network with a waste discharger terminal and a waste collector terminal according to another embodiment of the present invention includes a waste image transmitted from the waste discharger terminal and discharger information about the waste and the waste collection. Receiving information about a waste collector transmitted from a user terminal; Classifying the type of waste by analyzing the waste image or information on the discharger of the waste; Analyzing discharger information on the waste and calculating a service life of the waste; Analyzing the collector information on the waste to identify the appearance state of the waste; Determining whether the predicted value of the waste type obtained in the step of classifying the waste type matches the actual value of the waste type included in the discharger information or the collector information; And classifying the waste into a recycled product or a used product by using the information on the type of waste, the service life, and the appearance condition.

In this case, the step of classifying comprises classifying the waste as a recycled product or a used product if the predicted value and the actual value coincide in the determining step, and if the predicted value and the actual value do not coincide in the determining step The waste is classified as indeterminate or withholding judgment.

In this case, in the step of classifying, the classification result value is obtained using the information on the service life and the information on the appearance condition, and classified as used goods when the obtained classification result value is higher than a predetermined reference value. And, if it is lower than the reference value, the step of classifying as recycled products is performed.

According to various embodiments of the present invention, by analyzing the waste image or information on the discharger provided by the waste discharger for requesting disposal, the type of waste is automatically analyzed, and the waste is recycled or used according to the age and appearance of the waste. By automatically classifying them into goods, it has the effect of improving the economics of recycling waste resources,

By classifying waste into used or recycled products based on the discharger, collector, and accumulated experience data, not based on the subjective judgment of the waste collection company, the objectivity of the results is improved.

1 is a block diagram exemplarily illustrating the overall operation of an automatic sorting system for recycled or used goods according to an embodiment of the present invention;
FIG. 2 is a block diagram illustrating an embodiment of the automatic sorting system for recycled or used goods shown in FIG. 1;
3 is a view for explaining in more detail the automatic classification system for recycled or used goods shown in FIG. 2;
FIG. 4 is a block diagram for illustratively explaining an operation of the category classification unit shown in FIG. 3;
5 is a diagram illustrating a method of identifying a region of interest used by a distinguishing means according to an embodiment of the present invention;
6 is a diagram illustrating a process of discriminating a region of interest among waste images by the discriminating means of FIG. 4;
FIG. 7 is a diagram exemplarily illustrating an operation of the determination means shown in FIG. 4;
FIG. 8 is a block diagram for illustratively explaining the operation of the service life calculation unit shown in FIG. 3;
9 is a block diagram for illustratively explaining the operation of the appearance state identification unit shown in FIG. 3;
10 is a block diagram for illustratively explaining an operation of the processing method determination unit shown in FIG. 3;
11 is a flow chart for illustratively explaining each step of the automatic classification method of recycled or used goods according to another embodiment of the present invention, and,
12 is a flowchart for explaining another operation example of a method of automatically classifying recycled or used goods according to another embodiment of the present invention.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings, and various modifications may be implemented within the scope of the technical idea of the present invention.

1 is a block diagram illustrating the overall operation of an automatic sorting system for recycled or used goods according to an embodiment of the present invention. Referring to FIG. 1, an automatic waste sorting system 100 includes a communication device 110 and a sorting device 130. The communication device 110 may communicate with a plurality of waste discharger terminals 200-1 to 200-N and a plurality of waste collector terminals 300-1 to 300-N through a network in a wired or wireless manner. .

The communication device 110 receives a waste image and information about a waste discharger from the waste discharger terminal 200. In addition, the communication device 110 receives information about a waste collector from the waste collector terminal 300. Here, details of the waste image, discharger information, and collector information will be separately described below.

The classification device 130 may classify the discharged waste by type using the received waste image, discharger information, and collector information. Here, classification means classifying the waste into recycled products or used goods. Recyclables refer to goods that are sent to the process of obtaining necessary materials by breaking down waste to obtain raw materials that make up the waste. Used goods refer to goods that are sent to other users through a procedure that can be transferred for free or for free to perform the original function of waste.

The classification device 130 analyzes the waste image using a deep learning method, predicts the type of waste, calculates a predicted value of the type, and includes it in the discharger information received from the discharger terminal or the collector information received from the collector terminal. The actual value of the type of waste produced is obtained. The sorting device 130 compares whether the predicted value of the kind and the actual value of the kind match, and if there is a discrepancy, the next step of classifying the waste as recycled or used goods is not performed.

The classification device 130 calculates a classification result value by adding weights to the service life information and appearance condition information for the corresponding waste, and compares this with a reference value to calculate a classification result value only when the predicted value of the type and the actual value of the type coincide. Execute the process of classifying as used goods.

Alternatively, the sorting device 130 may classify the type of waste using the waste image photographed by the discharger and the discharger information input by the discharger. For example, waste images are classified using an image classification algorithm (image-based classification), and the discharger information input by the discharger is processed in natural language and the meaning is analyzed to classify the type (natural language processing-based classification). After that, the image-based classification and the natural language processing-based classification are compared with each other, and when the corresponding waste belongs to the same category, a process of automatically classifying it as a recycled or used product described below is performed.

FIG. 2 is a block diagram illustrating an embodiment of an automatic sorting system for recycled or used goods shown in FIG. 1. Referring to FIG. 2, the communication device 110 receives waste images, discharger information, and collector information from the waste discharger terminal 200 and the waste collector terminal 300 and stores it in a storage device (not shown) non-temporarily. do. The communication device 110 transmits the waste image, the discharger information, and the collector information to the classification device 130.

The classification device 130 may disperse and process received waste images, discharger information, and collector information into separate modules, respectively. Alternatively, these modules may be combined and configured to be processed as one module. Hereinafter, an example of processing each piece of information with a separate module will be described.

The category classification unit 131 analyzes the received waste image using a deep learning method to obtain identification information of the corresponding waste. A detailed description of this will be separately described below. The category classification unit 131 transmits the waste classification information (type predicted value) to the treatment method determination unit 137. The category classification unit 131 may classify the type of waste by grasping the meaning of the information input by the discharger through natural language processing. Alternatively, the category classification unit 131 may classify the type of waste using information on the type of waste input by the discharger. The category classification unit 131 may classify wastes discharged in various ways, such as image-based, natural language processing-based, and metadata-based, into furniture, home appliances, and the like.

The service life calculation unit 133 calculates the service life of the waste by using various information capable of calculating the use time, such as purchase time, manufacturing time, and product year, of the waste included in the discharger information. Here, the usage period means actual usage time according to the difference between the purchase time and the current time. For example, if the purchase time is May 2010 and the current time is June 2019, the actual usage time is calculated as 9 years and 1 month (109 months). The service life calculation unit 133 considers the actual use time, but finally calculates the service life information by reflecting the recommended use time of the product group to which the waste belongs. The service life calculation unit 133 transfers the final service life information to the processing method determination unit 137.

The appearance state identification unit 135 analyzes the information of the collector regarding the appearance state of the waste included in the collector information and calculates the appearance state information. Here, the appearance state means functional or aesthetic state information such as damage, scratches, and contamination of the exterior. For example, in the case of a three-tiered drawer, if there is a scratch of 10cm or more on the front of the three-tiered drawer, if the collector gives 60 out of 100 points according to the evaluation table, the score of 60 is calculated as information on the appearance condition. do. The appearance state identification unit 135 finally calculates appearance state information by comparing the feedback information accumulated by existing users as well as the collector's rating. The appearance state identification unit 135 transmits the final appearance state information to the processing method determination unit 137.

The treatment method determination unit 137 determines a treatment method of the corresponding waste using the waste identification information, the service life information, and the appearance state information. The treatment method determination unit 137 compares the predicted value of the corresponding waste included in the waste identification information with the actual value of the corresponding waste, and determines whether or not the treatment method according to whether or not they match. If the predicted value of the type and the actual value match, the corresponding waste is classified as recycled or used goods according to the service life information and the appearance condition information described below, and a treatment method is determined.

If the predicted value of the type and the actual value do not match, the judgment of the disposal method for the waste is not made. Alternatively, if the treatment method determination unit 137 cannot identify a specific product based on the predicted value (if the type of the product cannot be specified using the predicted value alone), Can be judged. Alternatively, when it is difficult to identify a specific item based on the predicted value (estimating a plurality of items only with the predicted value), the treatment method determination unit 137 may determine that the type of the waste is put on hold.

3 is a view for explaining in more detail the automatic classification system for recycled or used goods shown in FIG. 2. Referring to FIG. 3, a discharger A may take a picture through the discharger terminal while discharging a two-person table and a chair, and input related information to receive waste treatment. After receiving the received waste image and information on the waste, the recycled or used product automatic sorting system 100 of the present invention classifies which category the waste belongs to by the category sorting unit 131. For example, since the received waste image is "two-person table and chair", it is classified as a dining table by an image classification algorithm and largely classified as furniture. B discharger desktops are classified as PCs by the category classification unit 131 and largely classified as home appliances. The grand piano for which the discharger C has applied for disposal treatment is classified as a piano by the category classification unit 131, and is largely classified as other.

For the waste classified in this way, the age-of-use calculation unit 133 extracts information on the date of manufacture or the year of purchase for each major classified waste from the discharger information. For example, in the case of home appliances, the date of manufacture of the waste is extracted from the discharger information, the period of use up to the present point is calculated, and then a quantitative score is given according to the period of use. In the case of households or others, the year of purchase of the waste is extracted from the information on the discharger, the period of use up to the present point is calculated, and a quantitative score is given according to the period of use.

In the case of household appliance waste, the service life calculation unit 133 is 15 points if the usage period is less than 3 months, 10 points if less than 1 year, 7 points if less than 3 years, 3 points if less than 5 years, 1 point if less than 10 years, If it is more than that, a quantitative score is given as 0.

In the case of furniture or other waste, the service life calculation unit 133 is 15 points if the service period is less than 1 year, 10 points if less than 3 years, 7 points if less than 5 years, 3 points if less than 10 years, and 1 Points are given as quantitative scores for the waste.

After assigning a quantitative score according to the age of use of the waste by the age-of-use calculation unit 133, the appearance state identification unit 135 identifies damage and contamination based on the image of the waste, and if the number of damage is 0 15 points, less than 3, 10 points, less than 5, 7 points, less than 10, 3 points, and 10 or more, 1 point as a qualitative score. In addition, the appearance state identification unit 135 is 15 points if the contaminated area is 0%, 10 points if the range is 1% to 10%, 7 points if the range is 11% to 30%, and 3 points if the range is 30% to 50%, If it is 50% or more, 1 point is given as a qualitative score.

The points given by the service life calculation unit 133 and the appearance state identification unit 135 are summed to calculate a total score for each waste, and the total score is included in the waste information and transmitted to the treatment method determination unit 137.

The treatment method determination unit 137 analyzes the waste image and includes all items corresponding to the waste in the image, and if it matches the category classified by the category classification unit 131, assigns 2 points as a determination value, and If the image alone cannot be identified, 0 is given as the decision value, if there is a discrepancy, 0 is given as the decision value, and if the waste image is omitted, 0 is given as the decision value.

The treatment method determination unit 137 updates the total score to a final value by multiplying the total score included in the waste information by the determined value. The treatment method determination unit 137 compares the final value of the waste with a reference value, and if the final value is greater than the first reference value, classifies it as a used product, if it is less than the first reference value, it classifies it as a recycled product, and if it is less than the second reference value, the waste Classified as.

4 is a block diagram illustrating an operation of the category classifying unit illustrated in FIG. 2 by way of example. Referring to FIG. 4, the category classification unit 131 includes a discriminating means 1310 and a determining means 1312.

The discrimination means 1310 may distinguish between an ROI and an uninterested region in the input image using the Faster-RCNN method. More specifically, the discrimination means 1310 creates a bounding box for the region of interest including the waste in the waste image of an arbitrary size, thereby distinguishing the region of interest from the region of interest, and distinguishing the region of interest. Identify. Faster RCNN will be described separately below.

The determination means 1312 adjusts the region of interest to an appropriate size, and then classifies each type of waste by performing a classification function through a filter bank trained in advance for each type of waste. In the class classification block, the determination means 1312 converts data into a filter bank based on size adaptation for processing a bounding box of various sizes and information on a waste class detected in advance. A control block that performs the function of transmitting is provided. The Classification block can increase or decrease the number of classes according to the number of types of waste.

In addition, each class is pre-configured with a training data set consisting of a true positive and a false positive in order to distinguish the types of waste using binary classification. Is trained on. To this end, a separate second convolutional neural network configured by varying truth affirmation for each class can be used to identify the type of detailed items of waste.

In one embodiment of the present invention, the role of the control block is a function of extracting a bounding box image and a function of size adaptation of the extracted bounding box. In addition, when the control block is input to the CNN class corresponding to each type of waste in the bounding box image detected by the distinguishing unit 130, each CNN class determines what type of waste appears in the input image and is true. Outputs (True) or False value.

The distinguishing means 1310 includes an object classification module 1310a and a location identification module 1310b. The distinguishing means 1310 may distinguish between a region of interest and a region of non-interest in the received image using the Faster R-CNN method.

The object classification module 1310a of the discrimination means 1310 applies a sliding window to the first feature map and provides information on the existence of an object that determines whether or not an object is included in a specific group of pixels included in the sliding window. Generate. The location identification module 1310b of the discrimination means 1310 generates window location information capable of confirming the center location of the sliding window applied to the first feature map.

The operation of a network that proposes a region of interest using information on the existence of an object generated by the discrimination means 1310 and window location information will be separately described below.

5 is a diagram illustrating a method of identifying an ROI used by a discriminating means according to an embodiment of the present invention. Referring to FIG. 5, a sliding window is applied to a convolution feature map. K anchor boxes for the sliding window are created, and 2k object presence information for k anchor boxes are generated. In addition, 4k window position information is generated for k anchor boxes. 2k object presence information and 4k window position information are transmitted to the class layer and the regression layer, respectively, to identify the ROI through a convolutional neural network. A process of dividing an area containing waste into an area of interest in the waste-related image will be described below with reference to FIG. 5.

6 is a diagram illustrating a process of discriminating a region of interest among waste images by the discriminating means of FIG. 4. Referring to FIG. 6(a), the discrimination means 1310 largely divides the waste image into four areas. The first region (Region 1) is a region including most of the first waste, and the second region (Region 2) has similar characteristics to the first waste, but is not classified as a waste. The third region (Region 3) is a peripheral region other than the first waste, and the fourth region (Region 4) is a background region surrounding the first waste.

The distinguishing means 1310 may use Faster R-CNN to distinguish the first region and the second region excluding the third region and the fourth region constituting the waste image into a first region of interest (Fig. 6(b)). Reference). The distinguishing means 1310 may distinguish the first region of interest into the second region of interest by applying Faster R-CNN to the first region of interest (see FIG. 6(c)). That is, the first region of interest is composed of a first region and a second region, and the second region of interest is composed of only the first region. The first region and the second region share similar feature points, but since they are substantially different objects, the second region of interest may more accurately represent the waste type than the first region of interest. In other words, since the first region of interest includes the second region, the type, size, and weight of the waste can be distorted. Therefore, if the second region of interest that can more accurately extract objects is used, the type, size, and weight of the waste The back can be identified more accurately.

7 is a diagram exemplarily illustrating the operation of the determination means shown in FIG. 4. Referring to FIG. 6, the determination means 1312 receives a waste image including a region of interest (Region 1) as input data. The determination means 1312 changes the input data into a structure that can be input to the CNN in a size adaptation block. The determination means 1312 extracts features of the input data through a convolutional neural network learned for each waste class, and classifies them according to the extracted feature points, thereby identifying each type of wastes of more than 140 types.

Looking at the structure of a convolutional neural network for identifying the type of waste shown in FIG. 7, a first feature map is obtained by applying a first convolution to input data. The size is reduced by subsampling the first feature map, and at this time, a max pool is used. A second feature map is obtained by applying a second order convolution to image data having a smaller size. The size of the second feature map is reduced by performing sub-sampling, and a max pool is also used in this case. One-dimensional matrix values obtained after performing such feature extraction processing are mapped to the neural network, one at each input end. The one-dimensional matrix value is input to an input terminal of a neural network, and a result value classified by waste type is output as an output result.

FIG. 8 is a block diagram illustrating an example operation of the service life calculation unit shown in FIG. 2. Referring to FIG. 8, the service life calculation unit 133 includes an operation unit 1330 and a service life information DB 1332. The service life calculation unit 133 extracts start time information such as the purchase time, manufacturing time, and product year of the corresponding waste included in the discharger information received from the discharger terminal 200, and compares the extracted start time information with the current time. And calculate the service life.

For example, if the discharger information includes the time of purchase of waste X (June 2015: start time information) and the time of analysis of the expiration date (June 2019: current time) is checked, the use life of waste X is It is calculated as 4 years (48 months). Quantitative scores according to the service life of each product stored in the service life information DB 1332 are searched. Since the use period of X is 4 years, the calculation means 1330 gives 3 points as a quantitative score for waste X. If X is household or other, the calculation means 1330 gives 7 points as a quantitative score for waste X. That is, different quantitative scores may be given for each category of waste even during the same period of use.

FIG. 8 is a block diagram illustrating an operation of an external state identification unit shown in FIG. 2 by way of example. Referring to FIG. 8, the appearance state identification unit 135 includes an identification means 1350 and a score table DB 1352 for each appearance state. The appearance condition identification unit 135 searches the collector's rating regarding the appearance condition of the waste included in the collector information received from the collector terminal 300 from the appearance condition score table DB 1352 to qualitatively Calculate the score.

For example, if the collector's appearance condition rating included in the collector information is 3 damage and 10% contamination, the identification means 1350 is 7 points in the case of 3 in the damage score table, and the contamination range is 10%. In this case, it corresponds to 10 points, so 17 points are given as qualitative points according to the appearance condition.

10 is a block diagram illustrating an operation of a processing method determination unit illustrated in FIG. 2 by way of example. Referring to FIG. 10, the processing method determining unit 137 includes a data processing unit 1370 and a processing method determining unit 1372.

The data processing means 1370 receives service life information as quantitative information, and receives appearance state information as qualitative information. The data processing means 1370 calculates a classification result value by multiplying the quantitative information by a first weight and multiplying the qualitative information by a second weight. For example, if the age information as quantitative information is 25 points, the first weight is 1.1, the appearance status information as qualitative information is 17 points, and the second weight is 1.2, the classification result is 25×1.1 + 17×1.2 = 47.9 points. do.

The calculated result value is compared with a predefined reference value and classified as recycled or used goods according to whether the waste is large or small. For example, if the predetermined reference value is 50 points (the first reference value), wastes with 47.5 points are classified as recyclables, and wastes with 55 points are classified as used goods.

When the classification result value is calculated by the data processing means 1370, the treatment method determination means 1372 classifies the corresponding waste as recycled or used items based on the waste identification information and the classification result value. If the waste identification information does not match the predicted value and the actual value, it is possible to make a decision such as not being able to judge or withholding the judgment on the waste.

If the judgment is impossible, the predicted value of the waste type and the actual value do not match. For example, the category classification unit 131 analyzed the waste'X' based on the image by the CNN algorithm and predicted it as a "circular table" (the predicted value is a "circular table"), but the discharger terminal 200 or the collector terminal If the type of waste'X' included in the discharger information or the collector information received from (300) is "circular chair" (actual value'circular chair'), the actual value and the predicted value are inconsistent, so it is impossible to judge the waste Classify.

In the case of withholding judgment, there are a plurality of predicted values of the type of waste and partially coincide with the actual value. That is, the category classification unit 131 analyzed the'Y' waste based on the image by the CNN algorithm and predicted it as "dining table" or "office desk" (the predicted value is "dining table" or "desk"), and the discharger terminal 200 ) Or, if the type of'Y' waste included in the discharger information or the collector information received from the collector terminal 300 is "dining table" (actual value'dining table'), some of the predicted values and actual values coincide, but the rest Since they do not match, they are classified as pending judgment.

According to an embodiment of the present invention, if the type of waste identified using an image and the type of waste included in the discharger information or the collector information do not match or cannot be determined, the classification and evaluation of the waste is not performed. It is characterized by not. In this case, if the result of identifying the type of waste based on the image is different from the result recognized by the user, the degree of damage, damage, or deformation of the waste is severe. In this case, the waste is not classified as used goods. This is to improve the objectivity of classification as recycled or used goods.

11 is a flowchart illustrating each step of a method for automatically classifying waste according to another embodiment of the present invention. Referring to FIG. 11, the automatic waste classification method is a step of receiving a waste image and discharger information about the waste from the waste discharger terminal 200 (S1110), receiving information about the waste collector from the waste collector terminal 300 Step (S1120), the step of classifying the type of waste by recognizing the characteristic point pattern of the waste image using the CNN algorithm (S1130), the step of calculating the service life by analyzing the discharger information (S1140), the collector information Analyzing and calculating appearance state information (S1150), determining whether the predicted value for the type of waste (automatic identification by CNN) and the actual value (manual identification of the discharger or collector) match (S1160), the step ( If it matches in S1060), classification of the waste as a recycled product or used product using the information on the service life and appearance condition (S1170), if it does not match in step (S1160), the classification of the waste is not possible or withholding the judgment. It includes a step (S1180).

12 is a flowchart illustrating another operation example of a method for automatically classifying recycled or used goods according to another embodiment of the present invention. Referring to FIG. 12, referring to FIG. 12, the discharger inputs basic information (telephone number, discharge area, etc.), images of waste, and waste information (type, purchase date, manufacturing date, etc.) using a terminal. Execute the pre-processing step.

The automatic classification system 100 for recycled or used goods classifies a category for the corresponding waste by using basic information, waste images, and information on waste obtained in the pre-treatment step. The categories can be classified as household appliances, furniture or others. For categorized wastes, a quantitative score is given in consideration of the age of use (major step 1).

The automatic classification system for recycled or used goods 100 gives a qualitative score to the waste in consideration of the appearance state, using the appearance state information obtained from the collector, and adds the quantitative score and the qualitative score. The combined score is used to grade the waste. The administrator multiplies the sum score by the determined value and proceeds to classify the waste (major step 2).

If the total score multiplied by the determined value is greater than the reference value, the application for used goods is delivered to the discharger's terminal. If the total score multiplied by the determined value is less than the reference value, it is classified and processed as a recycled product (main step 3).

On the other hand, even if all the constituent elements constituting an embodiment of the present invention are described as being combined into one or operating in combination, the present invention is not necessarily limited to this embodiment. That is, within the scope of the object of the present invention, all of the constituent elements may be selectively combined and operated in one or more. In addition, although all the components may be implemented as one independent hardware, a program module that performs some or all functions combined in one or more hardware by selectively combining some or all of the components. It may be implemented as a computer program having Codes and code segments constituting the computer program may be easily inferred by those skilled in the art of the present invention. Such a computer program is stored in a non-transitory computer readable media that can be read by a computer and is read and executed by a computer, thereby implementing an embodiment of the present invention.

Here, the non-transitory readable recording medium is not a medium that stores data for a short moment, such as a register, cache, memory, etc., but a medium that stores data semi-permanently and can be read by a device. . Specifically, the above-described programs may be provided by being stored in a non-transitory readable recording medium such as a CD, DVD, hard disk, Blu-ray disk, USB, memory card, ROM, or the like.

In the above, preferred embodiments of the present invention have been illustrated and described, but the present invention is not limited to the specific embodiments described above, and is generally used in the technical field to which the present invention pertains without departing from the gist of the present invention claimed in the claims. Various modifications are possible by those skilled in the art of course, and these modifications should not be individually understood from the technical idea or perspective of the present invention.

100: automatic waste sorting system
110: communication device
130: classification device
131: Category Classifier
1310: means of distinction
1312: judgment means
133: service life calculation unit
1330: calculation means
1332: Use age information DB
135: appearance state identification unit
1350: means of identification
1352: User feedback information DB
137: processing method determination unit
1370: data processing means
1372: means for determining the treatment method

Claims (10)

In the system interconnected through a network with a waste discharger terminal and a waste collector terminal,
A communication device for receiving the waste image transmitted from the waste discharger terminal and discharger information about the waste, and receiving collector information about the waste transmitted from the waste collector terminal; And
Including; a classification device for classifying the discharged waste into recycled or used goods using the received waste image, the discharger information and the collector information,
The classification device includes a category classification unit for classifying the waste into furniture, household appliances, or other categories by analyzing the waste image, discharger information on the waste, and collector information on the waste, and analyzes discharger information on the waste. Then, a service life calculation unit that calculates the useful life of the waste, an appearance condition identification unit that identifies the appearance state of the waste by analyzing the collector information on the waste, and the type of the waste, the service life and the appearance state. Classification apparatus including a treatment method determination unit for multiplying the information by separate weights and summing them to calculate a classification result value, and comparing the calculated classification result value with a predetermined reference value to classify the corresponding waste as recycled or used goods; Include,
The category classification unit analyzes a waste image using a deep learning method to predict the type of waste to calculate a predicted value of the waste type, obtains an actual value of the waste type included in the discharger information and the collector information, An automatic classification system for recycled or used goods, characterized in that when the predicted value and the actual value are compared and matched with each other, the category of the corresponding waste is classified.
delete The method of claim 1,
The category classification unit compares the waste image information value included in the waste image with the reference image information value included in the plurality of machine-learned waste images to classify the waste into one of furniture, home appliances, or others. Automatic sorting system for recycled or used goods.
The method of claim 1,
The classification device, characterized in that the classification of the waste by matching categories by matching a result value obtained by processing the discharger information on the waste in natural language with a plurality of keyword values mapped to furniture, home appliances, or other categories. Automatic sorting system for recycled or used goods.
The method of claim 1,
The service life calculation unit is an automatic classification system for recycled products or used items, characterized in that the use of the information on the purchase time of the waste included in the discharger information and the recommended service life of the waste are calculated .
The method of claim 1,
The appearance state identification unit calculates appearance state information for the corresponding waste by using the score assigned according to the appearance state of the waste included in the collector information and the user's feedback information. Automatic sorting system.
delete In the automatic classification method using a system interconnected through a network with a waste discharger terminal and a waste collector terminal,
Receiving a waste image transmitted from the waste discharger terminal, discharger information about the waste, and collector information about the waste transmitted from the waste collector terminal;
By analyzing the waste image using a deep learning method, predicting the type of waste to calculate a predicted value of the waste type, obtaining the actual value of the waste type included in the discharger information and the collector information, and obtaining the predicted value and Comparing the actual values and classifying the type of the waste into households, household appliances, or other categories when they match each other;
Analyzing discharger information on the waste and calculating a service life of the waste;
Analyzing the collector information on the waste to identify the appearance state of the waste;
Determining whether the predicted value of the waste type obtained in the step of classifying the waste type matches the actual value of the waste type included in the discharger information or the collector information; And
The classification result value obtained by multiplying and summing the information on the type of waste, the age of use, and the appearance condition, respectively, by a separate weight, is calculated, and the calculated classification result value is compared with a predetermined reference value, and the corresponding waste is recycled or used. Automatic classification method of recycled or used goods, including the step of classifying as goods.
The method of claim 8,
In the step of classifying, if the predicted value and the actual value coincide in the determining step, the waste is classified as a recycled product or a used product, and if the predicted value and the actual value do not coincide in the determining step, the waste is Automatic classification method of recycled products or used goods, characterized in that the classification as judgment impossible or judgment hold.
The method of claim 9,
In the classifying step, a classification result value is obtained by using the information on the service life and the information on the appearance condition, and when the obtained classification result value is higher than a predetermined reference value, the classification is classified as used goods, and the reference value Automatic sorting method of recycled or used goods, characterized in that if it is lower than that classified as recycled goods.

Images