Waste sorting system considering characteristics of each waste sorting facility

Abstract

The present invention provides a waste sorting system considering the characteristics of each waste sorting facility and based on an artificial intelligence model trained to sort waste by using first training data based on first image information in an existing waste sorting facility. The waste sorting system comprises: a data preprocessing unit that corrects a first image by using second image information acquired in a new waste sorting facility and the first image information, and generates second training data; a model training unit that trains the artificial intelligence model by using the second training data; and a target object selection unit that selects a target object intended to be sorted from waste in the new waste sorting facility by using the artificial intelligence model with test data based on the second image information.

Description

A waste classification system that takes into account the characteristics of each waste classification facility.

[0001] The present invention relates to a waste classification system using artificial intelligence and robots, and more specifically, to a waste classification system that takes into account the characteristics of each waste classification business site, which reflect the different environmental characteristics of each waste classification business site.

[0002] A large amount of waste is produced every day from homes, factories, restaurants, etc., and many people separate and throw away food waste among general waste, recyclable waste, and food waste.

[0003] However, recyclable waste is still being thrown out mixed with general waste, and manpower is being directly used to separate it.

[0004] In addition, companies that use PET or glass bottles are importing recyclable waste from overseas because recyclable waste is not properly separated and is disposed of as general waste.

[0005] In Korea, various inventions are being made to solve these problems, and in particular, methods, devices, and systems for separating PET bottles are being applied for and registered as patents.

[0006] For example, Korean Patent Publication No. 10-2023-0084017, "Recyclable waste sorting device, method, and program using an artificial intelligence model," relates to a recyclable waste sorting device, which captures images of waste on a conveyor through at least one camera in a preset sorting area, analyzes the captured images based on an artificial intelligence model to determine the type of each waste on the conveyor, and analyzes the captured images using an artificial intelligence model to determine the type of waste for each waste on the conveyor.

[0007] Meanwhile, waste image data, the training data for AI models, can vary from one waste sorting facility to another due to environmental differences. This means that when an AI model trained on waste image data from Waste Sorting Facilities A is used to classify waste at Waste Sorting Facilities B, errors can occur.

[0008] The technical problem to be achieved by the present invention is to provide a waste classification system that takes into account the characteristics of each waste classification business site, which can utilize an artificial intelligence model learned from waste image data of existing waste classification businesses with different environmental differences in new waste classification businesses.

[0009] In order to solve such a problem, a waste classification system considering the characteristics of each waste classification business site according to an embodiment of the present invention is a waste classification system based on an artificial intelligence model trained to classify waste using first training data based on first image information from an existing waste classification business site, the system including: a data preprocessing unit for generating second training data by correcting the first image using second image information acquired from a new waste classification business site and the first image information; a model learning unit for learning the artificial intelligence model using the second training data; and a target object selection unit for selecting a target object to be classified from among waste from a new waste classification business site using the artificial intelligence model as test data based on the second image information.

[0010] Due to differences in the environment of the waste sorting facility and the performance of the sensor, the first waste image information and the second waste image information are different, and the image information may include color space, resolution, bit depth, metadata, density, and transmittance as components.

[0011] The above data preprocessing unit can generate second training data by correcting the first image using at least two components of the image information.

[0012] The first image correction and the artificial intelligence model retraining can be repeatedly performed until the classification error rate of the target object becomes lower than a preset standard through the artificial intelligence model.

[0013] After determining the importance of the components of the image information that affect the classification result of the target object through the above cyclical repetition execution, when applying the artificial intelligence model to another new waste classification facility, data can be preprocessed by giving priority to the components with high importance.

[0014] In addition to the technical problems of the present invention mentioned above, other features and advantages of the present invention are described below or may be clearly understood by a person skilled in the art to which the present invention pertains from such description and explanation.

[0015] According to the present invention as described above, the following effects are achieved.

[0016] The present invention can improve the waste classification performance of a new waste classification business by correcting the training data of an artificial intelligence model so that an artificial intelligence model used in an existing waste classification business with a different environment can be used in a new waste classification business.

[0017] In addition, other features and advantages of the present invention may be newly discovered through the embodiments of the present invention.

[0018] FIG. 1 is a schematic diagram of a waste classification system using an artificial intelligence-based robot arm to which a waste classification system that takes into account the characteristics of each waste classification business site according to one embodiment of the present invention is applied.

[0019] Figure 2 is a schematic diagram of a waste classification system that takes into account the characteristics of each waste classification business site according to one embodiment of the present invention.

[0020] Figure 3 is a drawing explaining a process of correcting first image information using first image information from an existing waste classification business operator and second image information from a new waste classification business operator.

[0021] FIG. 4 is a flowchart illustrating a process of classifying a target object through image correction according to one embodiment of the present invention.

[0022] FIG. 5 is a diagram illustrating a process of performing first image correction until the classification error rate of the target object becomes lower than a preset standard through cyclic iteration according to one embodiment of the present invention.

[0023] When adding reference numbers to components of each drawing in this specification, it should be noted that identical components are given the same numbers as much as possible even if they are shown in different drawings.

[0024] Meanwhile, the meanings of the terms described in this specification should be understood as follows.

[0025] Singular expressions should be understood to include plural expressions unless the context clearly defines otherwise, and terms such as “first”, “second”, etc. are intended to distinguish one component from another, and the scope of rights should not be limited by these terms.

[0026] The terms "include" or "have" should be understood as not excluding in advance the presence or addition of one or more other features, numbers, steps, operations, components or combinations thereof.

[0027] As used herein, the term "waste" includes plastics, PET bottles, glass bottles, glass, paper, Styrofoam, general waste, and industrial waste.

[0028] As used herein, the term "original image" refers to an image collected rather than generated by a generative model. Specifically, the original image may be an image collected by a sensor or an image input from an external source.

[0029] The target object used in this specification refers to an object to be classified among various types of objects such as waste.

[0030] Hereinafter, preferred embodiments of the present invention designed to solve the above problems will be described in detail with reference to the attached drawings.

[0031]

[0032] FIG. 1 is a schematic diagram of a waste classification system using an artificial intelligence-based robot arm to which a waste classification system that takes into account the characteristics of each waste classification business site according to one embodiment of the present invention is applied.

[0033] A waste classification system using an artificial intelligence-based robot arm may include a waste classification system (1000) that takes into account the characteristics of each business site according to the present invention, a vision camera (2000), and a robot arm (3000).

[0034] Objects (10), such as waste moving on a conveyor belt (4000), are collected through a vision camera (2000), and the objects to be classified are selected through a waste classification system (1000) that takes into account the characteristics of each business site according to the present invention, and then the objects can be classified through a robot arm (3000).

[0035] The waste classification system (1000) according to the present invention, which takes into account the characteristics of each business site, can select a target object to be classified from among waste (10) through an artificial intelligence model used by existing waste classification businesses.

[0036] Typically, when waste such as plastic, PET bottles, and glass bottles arrive at a waste sorting facility, they often arrive in distorted forms, such as crushed or broken, rather than in their entirety.

[0037] The AI ​​model used to classify waste in existing waste classification businesses uses waste images from existing waste classification businesses as training data, so the classification performance is good in existing waste classification businesses, but there is a problem that the classification performance is poor when the AI ​​model is applied to a new waste classification business with a different environment.

[0038] The waste classification system (1000) according to the present invention, which takes into account the characteristics of each waste classification business site, can improve the artificial intelligence model used in existing waste classification businesses with different environments so that it can be used in new waste classification businesses.

[0039] In addition to the vision camera (2000), various sensors can be used as equipment for collecting images of an object (10) used in the present invention. For example, an optical sensor, an infrared sensor, an ultrasonic sensor, a metal detector, an x-ray sensor, a radar sensor, etc. can be used.

[0040] The robot (3000) is a robot for classifying target objects moving on a conveyor belt, and not only suction-type robotic arms but also gripper-type robotic arms with various gripping forms can be used.

[0041] Figure 2 is a schematic diagram of a waste classification system that takes into account the characteristics of each waste classification business site according to one embodiment of the present invention.

[0042] Referring to FIG. 2, a waste classification system (1000) that takes into account the characteristics of each waste classification business site according to one embodiment of the present invention may include a data preprocessing unit (100), a model learning unit (200), and a target object selection unit (300).

[0043] The data preprocessing unit (100) can generate second training data by correcting the first image using first training data based on first image information from an existing waste classification facility and second image information acquired from a new waste classification facility.

[0044] Due to differences in the environment and sensor performance between the existing and new waste classification facilities, the first waste image information and the second waste image information are different.

[0045] At this time, image information may include color space, resolution, bit depth, metadata, density, and transmittance as components.

[0046] For example, the data preprocessing unit (100) can convert all images into a specific color space such as RGB or HSV and resize them to a fixed resolution to consistently correct the data.

[0047] Additionally, if the bit depth is different, the brightness or color information of the image may be different, so to compensate for this, the bit depth of the image can be matched to a specific value or normalization can be performed.

[0048] Additionally, the image's metadata can include information such as the shooting environment, camera settings, date and time, which can be used to determine what filters or corrections have been applied to the image and, if necessary, to perform reverse correction.

[0049] Additionally, since the brightness or transparency of the image may vary depending on the density or transmittance of the waste, this can be corrected using the image histogram equalization or algorithm.

[0050] Histogram equalization is a technique for enhancing the contrast of an image. It smooths the distribution of pixel values ​​in the image, making it more uniform overall. This improves the contrast of the image, making details more visible, especially in dark and bright areas.

[0051] Adaptive Histogram Equalization: Unlike basic histogram equalization, this method adjusts the contrast by considering the local area of ​​the image.

[0052] The data preprocessing unit (100) can use methods such as gain-offset correction, color correction, white balance correction, and gamma correction.

[0053] Gain-Offset Correction is a method of adjusting the brightness and contrast of an image by applying gain and offset values ​​to specific pixel values. Color Correction is a technique for correcting color distortion in an image that occurs for various reasons. White Balance Correction is a method for compensating for color changes that occur due to the characteristics of the camera sensor, lighting conditions, etc. Gamma Correction is a method of applying a nonlinear transformation to pixel values ​​to adjust the brightness and contrast of an image.

[0054] The data preprocessing unit (100) can correct the first image using at least two components of the image information to generate second training data.

[0055] Figure 3 is a drawing explaining a process of correcting first image information using first image information from an existing waste classification business operator and second image information from a new waste classification business operator.

[0056] Referring to FIG. 3, the first image (110) is an image acquired from an existing waste sorting facility, and the second image (120) is an image acquired from a new waste sorting facility. The data preprocessing unit (100) according to one embodiment of the present invention can generate second training data through a newly corrected image (130) by correcting the first image (110) using the first image (110) and the color space, resolution, and transmittance of the first image (110).

[0057] The model learning unit (200) can train an artificial intelligence model used in an existing waste classification facility using the second training data.

[0058] The target object selection unit (300) can select a target object to be classified using test data based on second image information acquired at a new waste classification facility using an artificial intelligence model learned using second training data.

[0059] Artificial intelligence models can use deep learning or neural networks. Neural network models can include various types of models such as CNN (Convolution Neural Network), R-CNN (Region with Convolution Neural Network), RPN (Region Proposal Network), RNN (Recurrent Neural Network), S-DNN (Stacking-based Deep Neural Network), S-SDNN (State-Space Dynamic Neural Network), Deconvolution Network, DBN (Deep Belief Network), RBM (Restrcted Boltzman Machine), Fully Convolutional Network, LSTM (Long Short-Term Memory) Network, Classification Network, etc. However, any artificial intelligence model that can identify the target object to be classified using waste image data is not limited to this.

[0060] FIG. 4 is a flowchart illustrating a process of classifying a target object through image correction according to one embodiment of the present invention, and FIG. 5 is a diagram illustrating a process of performing a first image correction until a classification error rate of a target object becomes lower than a preset standard through cyclic iteration according to one embodiment of the present invention.

[0061] Referring to FIGS. 4 and 5, a waste classification system (1000) that takes into account the characteristics of each waste classification business site according to an embodiment of the present invention can be iterated through a step (S100) of comparing a first image acquired from an existing waste classification business site with a second image acquired from a new waste classification business site, a step (S200) of a data preprocessing unit (100) comparing the first image and the second image to correct the first image to generate second training data, a step (S300) of retraining an artificial intelligence model used in an existing waste classification business site using the second training data through a classification model learning unit (200), a step (S400) of selecting a target object in a new waste classification business site using an artificial intelligence model through a target object selection unit (300), and a step (S500) of classifying the target object using a robot arm (300).

[0062] At this time, the cyclic repetition execution termination (S600) can be performed by cyclically repeating the first image correction and artificial intelligence model retraining until the classification error rate of the target object becomes lower than a preset standard.

[0063] If the classification error rate of the target object in the first iteration (literation 1) falls below a preset threshold, the second iteration (literation 2) is executed by modifying the combination of image information components. If the classification error rate exceeds the preset threshold through this iterative cycle, the previously used combination of image information components can be applied to another new waste classification facility, thereby reducing the number of iterations.

[0064] In addition, after determining the importance of the components of image information that affect the classification results of the target object through cyclical iteration, when applying the AI ​​model to another new waste classification facility, data preprocessing can be performed by giving priority to the components with high importance.

[0065] The systems or devices described above may be implemented as hardware components, or a combination of hardware components and software components. For example, the devices and components described in the embodiments may be implemented using one or more general-purpose computers or special-purpose computers, such as, for example, a processor, a controller, an arithmetic logic unit (ALU), a digital signal processor, a microcomputer, a field programmable gate array (FPGA), a programmable logic unit (PLU), a microprocessor, or any other device capable of executing instructions and responding to them. The processing device may execute an operating system (OS) and one or more software applications running on the operating system. The processing device may also access, store, manipulate, process, and generate data in response to the execution of the software. For ease of understanding, the processing device is sometimes described as being used alone; however, one of ordinary skill in the art will recognize that the processing device may include multiple processing elements and / or multiple types of processing elements. For example, a processing unit may include multiple processors, or a processor and a controller. Other processing configurations, such as parallel processors, are also possible.

[0066] Software may include a computer program, code, instructions, or a combination of one or more of these, which may configure a processing device to perform a desired operation or may independently or collectively command the processing device. The software and / or data may be embodied in any type of machine, component, physical device, virtual equipment, computer storage medium, or device for interpretation by the processing device or for providing instructions or data to the processing device. The software may also be distributed over networked computer systems and stored or executed in a distributed manner. The software and data may be stored on one or more computer-readable recording media.

[0067] The method according to the embodiment may be implemented in the form of program commands that can be executed through various computer means and recorded on a computer-readable medium. The computer-readable medium may include program commands, data files, data structures, etc., singly or in combination. The medium may continuously store a computer-executable program, or may temporarily store it for execution or download. In addition, the medium may be various recording means or storage means in the form of a single or multiple hardware combinations, and is not limited to a medium directly connected to a computer system, but may also be distributed over a network. Examples of the medium may include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical recording media such as CD-ROMs and DVDs, magneto-optical media such as floptical disks, and those configured to store program commands, including ROM, RAM, and flash memory. In addition, examples of other media may include recording media or storage media managed by app stores that distribute applications, sites that supply or distribute various software, servers, etc. Examples of program instructions include machine language code, such as that produced by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0068] It will be apparent to a person skilled in the art that the present invention described above is not limited to the above-described embodiments and the attached drawings, and that various substitutions, modifications, and changes are possible within the scope that does not depart from the technical spirit of the present invention.

Claims

1. In a waste classification system that considers the characteristics of each waste classification facility, based on an artificial intelligence model trained to classify waste using first training data based on first image information from an existing waste classification facility, A data preprocessing unit that generates second training data by correcting the first image using the second image information acquired at a new waste classification facility and the first image information; A model learning unit that trains the artificial intelligence model using the second training data; A waste classification system that considers the characteristics of each waste classification business site, including a target object selection unit that selects a target object to be classified from among wastes of a new waste classification business site using the artificial intelligence model as test data based on the second image information.

2. In paragraph 1, A waste classification system that considers the characteristics of each waste classification facility, wherein the first waste image information and the second waste image information are different due to differences in the environment and sensor performance of the waste classification facility, and the image information includes color space, resolution, bit depth, metadata, density, and transmittance as components.

3. In paragraph 2, A waste classification system that takes into account the characteristics of each waste classification business site, wherein the data preprocessing unit corrects the first image using at least two components of the image information to generate second training data.

4. In paragraph 3, A waste classification system that takes into account the characteristics of each waste classification business site, wherein the first image correction and the artificial intelligence model retraining are repeatedly performed until the classification error rate of the target object becomes below a preset standard through the artificial intelligence model.

5. In paragraph 4, A waste classification system that considers the characteristics of each waste classification facility, characterized in that, after determining the importance of the components of the image information that affect the classification result of the target object through the above cyclical repetition execution, when applying the artificial intelligence model to another new waste classification facility, data is preprocessed by giving priority to the components with high importance.

Images