CN112651318A - Image recognition-based garbage classification method, device and system - Google Patents
Image recognition-based garbage classification method, device and system Download PDFInfo
- Publication number
- CN112651318A CN112651318A CN202011510206.0A CN202011510206A CN112651318A CN 112651318 A CN112651318 A CN 112651318A CN 202011510206 A CN202011510206 A CN 202011510206A CN 112651318 A CN112651318 A CN 112651318A
- Authority
- CN
- China
- Prior art keywords
- target
- classified
- image
- garbage
- image recognition
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000010813 municipal solid waste Substances 0.000 title claims abstract description 127
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000001514 detection method Methods 0.000 claims abstract description 55
- 238000012549 training Methods 0.000 claims description 58
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 238000004806 packaging method and process Methods 0.000 claims description 5
- 230000000873 masking effect Effects 0.000 claims description 4
- 238000011897 real-time detection Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000010586 diagram Methods 0.000 description 6
- 239000002699 waste material Substances 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000007613 environmental effect Effects 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000009499 grossing Methods 0.000 description 3
- 239000010819 recyclable waste Substances 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 239000010806 kitchen waste Substances 0.000 description 2
- 238000013508 migration Methods 0.000 description 2
- 230000005012 migration Effects 0.000 description 2
- 230000001360 synchronised effect Effects 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000001960 triggered effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/10—Terrestrial scenes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a garbage classification method, a device and a system based on image recognition, wherein the method comprises the following steps: detecting whether a target to be classified exists in a preset area in real time; when the target to be classified exists, acquiring a target image of the target to be classified; inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified; according to the classification information, the target to be classified is put into a first target storage area; according to the invention, the garbage classification can be realized rapidly and accurately through the image recognition technology, so that a user can be helped to perform garbage classification treatment rapidly and conveniently, the accuracy of garbage classification is improved, and the workload of secondary garbage sorting is reduced; the garbage classified and identified is automatically thrown into the corresponding garbage storage area, manual throwing of a user is replaced, intelligent operation of classified throwing of the garbage is achieved, and experience of the user is improved.
Description
Technical Field
The invention relates to the technical field of image recognition, in particular to a garbage classification method, device and system based on image recognition.
Background
The garbage classification is one of the bottlenecks which currently restrict the development of the environmental protection business of China, and is also one of the sources which cause environmental pollution and difficult resource recycling. In recent years, the work of garbage classification is accelerated in China, families are the source of household garbage, and the current quartering method classification mode is complex, the participation of residents is low, and the work development of garbage classification is not facilitated. The current garbage classification is distinguished by the consciousness of people, and people lack comprehensive knowledge of garbage classification at present, so that the people cannot accurately remember and judge which type the complicated and various garbage belongs to, the garbage can not be accurately classified and put in, and great pressure is caused to urban environment and garbage treatment.
It can be seen that how to realize the rapid and accurate classified delivery of garbage in daily life is a problem to be solved urgently at present.
Disclosure of Invention
Aiming at the defects in the prior art, the garbage classification method, the garbage classification device and the garbage classification system based on image recognition solve the problem that the domestic garbage is not accurately thrown in the prior art, realize automatic classification and automatic throwing of garbage and improve the accuracy of garbage classification.
In a first aspect, the present invention provides a method for classifying garbage based on image recognition, wherein the method includes: detecting whether a target to be classified exists in a preset area in real time; when the target to be classified exists, acquiring a target image of the target to be classified; inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified; and according to the classification information, the target to be classified is put into a first target storage area.
Optionally, before inputting the target image to a target detection model for image recognition, the method further includes: acquiring a sample image set; marking the sample image set with data to generate an annotation data set, wherein the annotation data set comprises a training image set and annotation data matched with the training image set; performing data enhancement on the labeled data set to obtain a training data set; and inputting the training data set into an artificial neural network for iterative training according to a label smooth loss function to obtain the target detection model.
Optionally, performing data enhancement on the labeled data set to obtain a training data set, including: acquiring a target detection area of a current training image in the marked data set according to a first preset rule; replacing pixels of the target detection area according to preset pixels to obtain a first sample image; turning over the current training image according to a second preset rule to obtain a second sample image; constructing a target sample image by weighting a linear difference value according to the first sample image and the second sample image; and fusing all the training images and all the target sample images to obtain the training data set.
Optionally, detecting whether the preset area has the target to be classified in real time includes: sending a first detection signal to the preset area; receiving a second detection signal after the first detection signal is reflected; acquiring a target distance between the target and a reflection point according to the first detection signal and the second detection signal; and comparing the target distance with a preset distance, and judging whether the target to be classified exists in the preset area.
Optionally, when the target to be classified exists, acquiring a target image of the target to be classified, including: when the target distance is smaller than the preset distance, determining that the target to be classified exists in the preset area; collecting an original image of the target to be classified; and cutting and masking the original image to obtain the target image.
Optionally, the step of placing the target to be classified into a first target storage area according to the classification information includes: according to the classification information, position information of a first target storage area corresponding to the classification information is obtained; rotating the preset area to be right above the first target storage area according to the position information; and throwing the target to be classified on the preset area to the first target storage area.
Optionally, before detecting whether the target to be classified exists in the preset area in real time, the method further includes: acquiring the current storage capacity of each storage area in real time; when the current storage capacity is larger than or equal to a preset storage capacity, acquiring a second target storage area corresponding to the current storage capacity; and sending out a voice prompt according to the second target storage area.
Optionally, when the current storage capacity is greater than or equal to a preset storage capacity, after a second target storage area corresponding to the current storage capacity is obtained, the method further includes: and packaging and transferring the objects in the second target storage area, and releasing the storage capacity of the second target storage area.
In a second aspect, the present invention provides a garbage classification device based on image recognition, the device comprising: the detection module is used for detecting whether the preset area has the target to be classified in real time; the image acquisition module is used for acquiring a target image of the target to be classified when the target to be classified exists; the image recognition module is used for inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified; and the delivery module is used for delivering the target to be classified to a first target storage area according to the classification information.
In a third aspect, the invention provides a garbage classification system based on image recognition, which includes the above garbage classification device based on image recognition.
Compared with the prior art, the invention has the following beneficial effects:
according to the invention, the garbage classification can be realized rapidly and accurately through the image recognition technology, so that a user can be helped to perform garbage classification treatment rapidly and conveniently, the accuracy of garbage classification is improved, and the workload of secondary garbage sorting is reduced; the garbage classified and identified is automatically thrown into the corresponding garbage storage area, manual throwing of a user is replaced, intelligent operation of classified throwing of the garbage is achieved, and experience of the user is improved.
Drawings
Fig. 1 is a schematic flowchart illustrating a garbage classification method based on image recognition according to an embodiment of the present invention;
fig. 2 is a schematic flowchart of another image recognition-based garbage classification method according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a garbage classification device based on image recognition according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a garbage classification system based on image recognition according to an embodiment of the present invention;
fig. 5 is a schematic working diagram of a garbage classification system based on image recognition according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some embodiments of the present application, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
Fig. 1 is a schematic flowchart illustrating a garbage classification method based on image recognition according to an embodiment of the present invention; as shown in fig. 1, the image recognition-based garbage classification method specifically includes the following steps:
step S101, detecting whether a preset area has a target to be classified in real time;
specifically, detecting whether a target to be classified exists in a preset area in real time includes: sending a first detection signal to the preset area; receiving a second detection signal after the first detection signal is reflected; acquiring a target distance between the target and a reflection point according to the first detection signal and the second detection signal; and comparing the target distance with a preset distance, and judging whether the target to be classified exists in the preset area.
It should be noted that the preset area may be a garbage placing plate disposed below the garbage can lid, a plurality of microwave radars are disposed right above the garbage placing plate, the microwave radars transmit electromagnetic wave signals to the garbage placing plate in real time, when the electromagnetic wave signals are reflected by objects on the garbage placing plate or the garbage placing plate, echo signals are formed, according to a time difference between transmission of the electromagnetic wave signals and reception of the echo signals, a target distance between the microwave radars and a reflection point is calculated, the target distance is compared with the preset distance, when the target distance is equal to the preset distance, it is determined that there is no object to be classified in the preset area, and when the target distance is smaller than the preset distance, it is determined that there is an object to be classified in the preset area.
When an object to be classified exists in the preset area, an electromagnetic wave signal transmitted by the microwave radar is reflected by the object to be classified, so that the calculated distance from the microwave radar to a reflection point is smaller than the distance from the microwave radar to the preset area; the first detection signal in this embodiment is the electromagnetic wave signal, and the second detection signal is the echo signal.
In another embodiment of the present invention, whether the target to be classified exists in the preset area may also be detected by an infrared device.
Step S102, when the target to be classified exists, acquiring a target image of the target to be classified;
in this embodiment, when the target to be classified exists, acquiring a target image of the target to be classified includes: when the target distance is smaller than the preset distance, determining that the target to be classified exists in the preset area; collecting an original image of the target to be classified; and cutting and masking the original image to obtain the target image.
Specifically, when the target to be classified is determined to exist in the preset area, an image acquisition device arranged above the preset area is triggered to acquire an original image of the target to be classified, the original image is cut, and the related area of the garbage placing plate is masked in the image, so that the image only has the area of the target to be classified, the data volume of image processing is reduced, and the efficiency of image identification is improved.
Step S103, inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified;
in this embodiment, classification information includes, but is not limited to, kitchen waste, harmful waste, recyclable waste, and other waste, and the target image including the target to be classified is input into a trained target detection model for image recognition, so as to obtain specific classification information of the target to be classified, for example, the classification information of the target to be classified is recyclable waste.
And step S104, according to the classification information, the target to be classified is put into a first target storage area.
Specifically, the step of placing the target to be classified into a first target storage area according to the classification information includes: according to the classification information, position information of a first target storage area corresponding to the classification information is obtained; rotating the preset area to be right above the first target storage area according to the position information; and throwing the target to be classified on the preset area to the first target storage area.
In this embodiment, since the classification information includes four types of kitchen waste, harmful waste, recyclable waste, and other waste, the waste storage area also includes the four types described above, the four waste storage areas are independently disposed in different areas of the trash can, and correspond to different position information, when the classification information of the object to be classified is determined, the corresponding storage area of the object to be classified, that is, the first object storage area, can be obtained, the preset area is prevented from being moved to a position directly above the first object storage area by rotation or movement, and the object to be classified is thrown into the first object storage area.
Compared with the prior art, the invention has the following beneficial effects:
according to the invention, the garbage classification can be realized rapidly and accurately through the image recognition technology, so that a user can be helped to perform garbage classification treatment rapidly and conveniently, the accuracy of garbage classification is improved, and the workload of secondary garbage sorting is reduced; the garbage classified and identified is automatically thrown into the corresponding garbage storage area, manual throwing of a user is replaced, intelligent operation of classified throwing of the garbage is achieved, and experience of the user is improved.
Fig. 2 is a schematic flowchart of another image recognition-based garbage classification method according to an embodiment of the present invention; as shown in fig. 2, before the target image is input to the target detection model for image recognition, the method provided by the present invention further includes the following steps:
step S201, obtaining a sample image set;
step S202, marking the sample image set with data to generate a marked data set, wherein the marked data set comprises a training image set and marked data matched with the training image set;
step S203, performing data enhancement on the labeled data set to obtain a training data set;
and S204, inputting the training data set into an artificial neural network for iterative training according to a label smooth loss function to obtain the target detection model.
Further, performing data enhancement on the labeled data set to obtain a training data set, including: acquiring a first target storage area of the current training image in the marked data set according to a first preset rule; replacing pixels of the first target storage area according to preset pixels to obtain a first sample image; turning over the current training image according to a second preset rule to obtain a second sample image; constructing a target sample image by weighting a linear difference value according to the first sample image and the second sample image; and fusing all the training images and all the target sample images to obtain the training data set.
In this embodiment, the training sample data is subjected to marking processing to generate the jpg and txt, further, batches are randomly acquired, the complete paths of the pictures are spliced according to the comparison table, the pictures and the label list are acquired, label category processing is performed, and further, the labels are smoothed to prevent overfitting. At each iteration, (xi, yi) is not put directly into the training set, but an error rate epsilon is set, and (xi, yi) is substituted into the training with a probability of 1-epsilon, and (xi,1-yi) is substituted with a probability of epsilon. Label smoothing improves the final accuracy. Furthermore, data enhancement is carried out, and the characteristics of the contour, the texture, the position distribution and the like of various resolution, multi-scale and multi-granularity pictures can be captured more effectively. The data enhancement uses mixed training, random erasing and random inversion, and the mixed training mode is to randomly take two samples and construct a new sample through weighted linear interpolation. The random erasing is to randomly select a rectangular area in the original image, replace the pixels of the area with random values, and shield the pictures participating in training to different degrees. Random flipping is the random horizontal/vertical flipping of training data.
In another embodiment of the present invention, before detecting whether there is an object to be classified in a preset area in real time, the method further includes: acquiring the current storage capacity of each storage area in real time; when the current storage capacity is larger than or equal to a preset storage capacity, acquiring a second target storage area corresponding to the current storage capacity; and sending out a voice prompt according to the second target storage area.
Further, when the current storage capacity is greater than or equal to a preset storage capacity, after a second target storage area corresponding to the current storage capacity is obtained, the method further includes: and packaging and transferring the objects in the second target storage area, and releasing the storage capacity of the second target storage area.
It should be noted that, whether each storage area is full is detected in real time through an infrared detection device or a microwave radar, and when the storage area is full, a voice prompt can be sent out to remind a user to pack and clean garbage; the garbage in the area can be automatically packaged, and the packaged garbage is transferred out of the garbage storage area in a pushing mode, so that the garbage can be automatically classified and automatically thrown in after the storage capacity of the storage area is released.
Fig. 3 is a schematic structural diagram of a garbage classification device based on image recognition according to an embodiment of the present invention; as shown in fig. 3, the image recognition-based garbage classification apparatus provided by the present invention specifically includes:
the detection module 310 is configured to detect whether a target to be classified exists in a preset area in real time;
the image acquisition module 320 is used for acquiring a target image of the target to be classified when the target to be classified exists;
the image recognition module 330 is configured to input the target image into a target detection model for image recognition, so as to obtain classification information of the target to be classified;
and the delivery module 340 is configured to deliver the target to be classified to a first target storage area according to the classification information.
In this embodiment, the apparatus further includes: the sample image acquisition module is used for acquiring a sample image set; the data marking module is used for marking data of the sample image set to generate a marked data set, and the marked data set comprises a training image set and marked data matched with the training image set; the data enhancement module is used for enhancing the data of the labeled data set to obtain a training data set; and the training module is used for inputting the training data set into an artificial neural network for iterative training according to the label smooth loss function to obtain the target detection model.
Wherein the data enhancement module comprises: the target detection area acquisition module is used for acquiring a target detection area of the current training image in the annotation data set according to a first preset rule; the replacing module is used for replacing the pixels of the target detection area according to preset pixels to obtain a first sample image; the overturning module is used for overturning the current training image according to a second preset rule to obtain a second sample image; a construction module for constructing a target sample image by weighting a linear difference value according to the first sample image and the second sample image; and the fusion module is used for fusing all the training images and all the target sample images to obtain the training data set.
In this embodiment, the detecting module 310 includes: the signal sending module is used for sending a first detection signal to the preset area; the signal receiving module is used for receiving a second detection signal after the first detection signal is reflected; the target distance acquisition module is used for acquiring a target distance between the reflection point and the target according to the first detection signal and the second detection signal; and the judging module is used for comparing the target distance with a preset distance and judging whether the target to be classified exists in the preset area.
In this embodiment, the image capturing module 320 includes: the determining module is used for determining that the target to be classified exists in the preset area when the target distance is smaller than the preset distance; the acquisition module is used for acquiring an original image of the target to be classified; and the cutting module is used for cutting and masking the original image to obtain the target image.
In this embodiment, the delivering module 340 includes: the position information module is used for obtaining position information of the first target storage area corresponding to the classification information according to the classification information; the rotating module is used for rotating the preset area to be right above the first target storage area according to the position information; and the placing module is used for throwing the target to be classified on the preset area to the first target storage area.
In this embodiment, the apparatus further includes: the storage capacity acquisition module is used for acquiring the current storage capacity of each storage area in real time; the second target storage area acquisition module is used for acquiring a second target storage area corresponding to the current storage capacity when the current storage capacity is larger than or equal to the preset storage capacity; and the voice prompt module is used for sending out voice prompts according to the second target storage area.
In this embodiment, the apparatus further includes: and the packaging module is used for packaging and transferring the objects in the second target storage area and releasing the storage capacity of the second target storage area.
Fig. 4 is a schematic structural diagram of a garbage classification system based on image recognition according to an embodiment of the present invention; as shown in fig. 4, the image recognition-based garbage classification system provided by this embodiment includes intelligent classification garbage bin and server, the intelligent classification garbage bin includes the bucket head, the staving, collapsible baffle, and the equipment is born to the garbage classification rubbish (environmental protection plastic refuse bag), and vision recognition module, infrared detection module, speech recognition module, little circuit control panel, little circuit communication module, 360 degrees steering wheel, step motor. The bucket head contains visual identification module, infrared detection module, collapsible baffle, and the staving contains 4 rubbish charge space, and the charge space uses plastic baffle to separate, and speech recognition module is equipped with at the avris of staving to carry environmental protection plastic refuse bag that can be convenient, microcircuit accuse plate, microcircuit communication module, 360 degrees steering wheel. The voice recognition module comprises a sound pick-up, a broadcast device and a voice recognition SDK module.
The garbage can cover is composed of a can cover and an infrared detection device, when the can cover is opened, the infrared detection device is started to detect whether the garbage containing device is full, if the garbage containing device is full, the voice recognition module is used for broadcasting that the garbage is full, otherwise, whether garbage is placed on the garbage placing plate on the garbage can is detected, and if the garbage is placed, an instruction is sent to the garbage can body. Otherwise, the detection is continued after waiting for 5 seconds or the garbage can head is closed. The garbage can body comprises four garbage containing devices, such as an ordinary garbage can and plastic garbage bags, and is convenient for garbage classification quartering method collection and temporary storage. The image acquisition system comprises a camera and a light source, can acquire clear images on the garbage placing plate and transmits the images to the garbage classification and identification system through the Wi-Fi module.
Further, the resolution of the camera is determined by the garbage type and the shooting field size; the light source is a white LED mesopore backlight source, which is beneficial to clear shooting; further, after the garbage classification and identification system identifies the garbage, the lower computer is informed to start the stepping motor and the steering engine through the Wi-Fi module to start the work. Further, the steering engine rotates to a specified angle to wait after receiving the instruction, and further, the stepping motor receives the instruction to control the opening of the garbage placing plate; so that the garbage falls into the designated garbage accommodating device through the garbage placing plate; further, after the garbage placing plate is opened for 2 seconds, the lower computer sends an instruction to the stepping motor to close the garbage placing plate. Further, the voice broadcasting module plays the classification of the garbage.
Fig. 5 is a schematic working diagram of a garbage classification system based on image recognition according to an embodiment of the present invention, which includes the specific steps of:
step 1: the garbage placing plate of the garbage can body is used for placing garbage and recognizing the garbage by the infrared sensor, turning on the light source and informing the lower computer that the Wi-Fi module is initialized and ready for acquiring images;
step 2: the image acquisition system photographs the garbage on the garbage placing plate and uploads the photographed garbage to the garbage classification and identification system through the Wi-Fi module;
and step 3: the garbage classification recognition system receives and cuts the pictures, and performs frosting processing on relevant areas of the garbage placing plate in the pictures to enable the pictures to be only garbage areas. The algorithm adopted by the identification system is an artificial neural network, the migration model is EfficientNetB5, and a large amount of sample garbage classification quartering method identification training is carried out before the identification system is used. The algorithm will continue to be tuned and trained after being put into use.
And 3.1, reading and preprocessing data, marking training sample data to generate the jpg and txt, further randomly acquiring batches, splicing the complete paths of the pictures according to a comparison table, acquiring the pictures and a label list, carrying out label type processing, and further smoothing the labels to prevent overfitting. At each iteration, (xi, yi) is not put directly into the training set, but an error rate epsilon is set, and (xi, yi) is substituted into the training with a probability of 1-epsilon, and (xi,1-yi) is substituted with a probability of epsilon. Label smoothing improves the final accuracy. Furthermore, data enhancement is carried out, and the characteristics of the contour, the texture, the position distribution and the like of various resolution, multi-scale and multi-granularity pictures can be captured more effectively. The data enhancement uses mixed training, random erasing and random inversion, and the mixed training mode is to randomly take two samples and construct a new sample through weighted linear interpolation. The random erasing is to randomly select a rectangular area in the original image, replace the pixels of the area with random values, and shield the pictures participating in training to different degrees. Random flipping is the random horizontal/vertical flipping of training data.
And 3.2, realizing a model network structure, wherein the migration learning model adopted by the system is EfficientNetB5, training sample data is 200 thousands, the maximum output types are 124, the data in each batch is 3000 times, the data is thrown for 660 times, the basic learning rate is 0.00001, and the iteration times are 60 times.
And 3.3, after the training is finished, deploying the training on a server and providing an API for an external interface to use.
And 4, step 4: and the garbage classification and identification system returns the first 5 maximum probability values through an API (application program interface), if all the maximum probability values belong to a certain large class, the large class is returned, and if all the maximum probability values do not belong to the certain large class, the large class with the highest probability is returned. And informing the lower computer through the Wi-Fi module.
In another embodiment of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the following steps when executing the computer program: detecting whether a target to be classified exists in a preset area in real time; when the target to be classified exists, acquiring a target image of the target to be classified; inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified; and according to the classification information, the target to be classified is put into a first target storage area.
In a further embodiment of the invention, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, realizes the following steps: detecting whether a target to be classified exists in a preset area in real time; when the target to be classified exists, acquiring a target image of the target to be classified; inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified; and according to the classification information, the target to be classified is put into a first target storage area.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by a computer program, which can be stored in a non-volatile computer-readable storage medium, and can include the processes of the embodiments of the methods described above when the program is executed. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
It is noted that, in this document, relational terms such as "first" and "second," and the like, may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
Claims (10)
1. A method for classifying garbage based on image recognition is characterized in that the method comprises the following steps:
detecting whether a target to be classified exists in a preset area in real time;
when the target to be classified exists, acquiring a target image of the target to be classified;
inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified;
and according to the classification information, the target to be classified is put into a first target storage area.
2. The image recognition-based spam classification method of claim 1 wherein prior to inputting the target image into a target detection model for image recognition, the method further comprises:
acquiring a sample image set;
marking the sample image set with data to generate an annotation data set, wherein the annotation data set comprises a training image set and annotation data matched with the training image set;
performing data enhancement on the labeled data set to obtain a training data set;
and inputting the training data set into an artificial neural network for iterative training according to a label smooth loss function to obtain the target detection model.
3. The image recognition-based garbage classification method of claim 2, wherein the data enhancement of the labeled data set to obtain a training data set comprises:
acquiring a target detection area of a current training image in the marked data set according to a first preset rule;
replacing pixels of the target detection area according to preset pixels to obtain a first sample image;
turning over the current training image according to a second preset rule to obtain a second sample image;
constructing a target sample image by weighting a linear difference value according to the first sample image and the second sample image;
and fusing all the training images and all the target sample images to obtain the training data set.
4. The image recognition-based garbage classification method according to claim 1, wherein the real-time detection of whether the preset area has the target to be classified comprises:
sending a first detection signal to the preset area;
receiving a second detection signal after the first detection signal is reflected;
acquiring a target distance between the target and a reflection point according to the first detection signal and the second detection signal;
and comparing the target distance with a preset distance, and judging whether the target to be classified exists in the preset area.
5. The image recognition-based garbage classification method according to claim 4, wherein when the object to be classified exists, acquiring a target image of the object to be classified comprises:
when the target distance is smaller than the preset distance, determining that the target to be classified exists in the preset area;
collecting an original image of the target to be classified;
and cutting and masking the original image to obtain the target image.
6. The image recognition-based trash classification method of claim 1, wherein the throwing the objects to be classified into a first object storage area according to the classification information comprises:
according to the classification information, position information of a first target storage area corresponding to the classification information is obtained;
rotating the preset area to be right above the first target storage area according to the position information;
and throwing the target to be classified on the preset area to the first target storage area.
7. The image recognition-based garbage classification method according to any one of claims 1-6, wherein before detecting whether the target to be classified exists in the preset area in real time, the method further comprises:
acquiring the current storage capacity of each storage area in real time;
when the current storage capacity is larger than or equal to a preset storage capacity, acquiring a second target storage area corresponding to the current storage capacity;
and sending out a voice prompt according to the second target storage area.
8. The image recognition-based trash classification method of claim 7, wherein when the current storage capacity is greater than or equal to a preset storage capacity, after acquiring a second target storage area corresponding to the current storage capacity, the method further comprises:
and packaging and transferring the objects in the second target storage area, and releasing the storage capacity of the second target storage area.
9. An image recognition-based garbage classification device, characterized in that the device comprises:
the detection module is used for detecting whether the preset area has the target to be classified in real time;
the image acquisition module is used for acquiring a target image of the target to be classified when the target to be classified exists;
the image recognition module is used for inputting the target image into a target detection model for image recognition to obtain the classification information of the target to be classified;
and the delivery module is used for delivering the target to be classified to a first target storage area according to the classification information.
10. An image recognition based garbage classification system, characterized in that the garbage classification system comprises the image recognition based garbage classification device of claim 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011510206.0A CN112651318A (en) | 2020-12-19 | 2020-12-19 | Image recognition-based garbage classification method, device and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011510206.0A CN112651318A (en) | 2020-12-19 | 2020-12-19 | Image recognition-based garbage classification method, device and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112651318A true CN112651318A (en) | 2021-04-13 |
Family
ID=75355174
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011510206.0A Pending CN112651318A (en) | 2020-12-19 | 2020-12-19 | Image recognition-based garbage classification method, device and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112651318A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114077877A (en) * | 2022-01-19 | 2022-02-22 | 人民中科(济南)智能技术有限公司 | Newly added garbage identification method and device, computer equipment and storage medium |
| CN114219010A (en) * | 2021-11-24 | 2022-03-22 | 北京华能新锐控制技术有限公司 | Garbage classification method based on prior guide antagonism characterization learning |
| CN116342895A (en) * | 2023-05-31 | 2023-06-27 | 浙江联运知慧科技有限公司 | Method and system for improving sorting efficiency of renewable resources based on AI (advanced technology attachment) processing |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107054936A (en) * | 2017-03-23 | 2017-08-18 | 广东数相智能科技有限公司 | A kind of refuse classification prompting dustbin and system based on image recognition |
| WO2019000929A1 (en) * | 2017-06-30 | 2019-01-03 | 京东方科技集团股份有限公司 | Garbage sorting and recycling method, garbage sorting equipment, and garbage sorting and recycling system |
| CN110271792A (en) * | 2019-06-24 | 2019-09-24 | 浙江云澎科技有限公司 | A kind of dustbin of the automatic sorting rubbish of energy |
| CN110550347A (en) * | 2019-08-23 | 2019-12-10 | 珠海格力电器股份有限公司 | Garbage classification recycling method, user terminal and garbage can |
| CN110861852A (en) * | 2019-11-28 | 2020-03-06 | 石家庄邮电职业技术学院(中国邮政集团公司培训中心) | Automatic classification garbage can based on visual identification and classification method |
| CN111832607A (en) * | 2020-05-28 | 2020-10-27 | 东南大学 | Bridge disease real-time detection method based on model pruning |
-
2020
- 2020-12-19 CN CN202011510206.0A patent/CN112651318A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107054936A (en) * | 2017-03-23 | 2017-08-18 | 广东数相智能科技有限公司 | A kind of refuse classification prompting dustbin and system based on image recognition |
| WO2019000929A1 (en) * | 2017-06-30 | 2019-01-03 | 京东方科技集团股份有限公司 | Garbage sorting and recycling method, garbage sorting equipment, and garbage sorting and recycling system |
| CN110271792A (en) * | 2019-06-24 | 2019-09-24 | 浙江云澎科技有限公司 | A kind of dustbin of the automatic sorting rubbish of energy |
| CN110550347A (en) * | 2019-08-23 | 2019-12-10 | 珠海格力电器股份有限公司 | Garbage classification recycling method, user terminal and garbage can |
| CN110861852A (en) * | 2019-11-28 | 2020-03-06 | 石家庄邮电职业技术学院(中国邮政集团公司培训中心) | Automatic classification garbage can based on visual identification and classification method |
| CN111832607A (en) * | 2020-05-28 | 2020-10-27 | 东南大学 | Bridge disease real-time detection method based on model pruning |
Non-Patent Citations (2)
| Title |
|---|
| CHRISTIAN SZEGEDY GOOGLE INC等: "Rethinking the Inception Architecture for Computer Vision", 2016 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, 31 December 2016 (2016-12-31), pages 2818 - 2626 * |
| 言有三: "一文道尽深度学习中的数据增强方法", pages 2 - 3, Retrieved from the Internet <URL:https://blog.csdn.net/hacker_long/article/details/100137933?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522170615352616777224440809%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fall.%2522%257D&request_id=170615352616777224440809&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2~all~first_rank_ecpm_v1~rank_v31_ecpm-4-100137933-null-null.142^v99^pc_search_result_base5&utm_term=%E4%B8%80%E6%96%87%E9%81%93%E5%B0%BD%E6%B7%B1%E5%BA%A6%E5%AD%A6%E4%B9%A0%E4%B8%AD%E7%9A%84%E6%95%B0%E6%8D%AE%E5%A2%9E%E5%BC%BA%E6%96%B9%E6%B3%95&spm=1018.2226.3001.4449> * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114219010A (en) * | 2021-11-24 | 2022-03-22 | 北京华能新锐控制技术有限公司 | Garbage classification method based on prior guide antagonism characterization learning |
| CN114077877A (en) * | 2022-01-19 | 2022-02-22 | 人民中科(济南)智能技术有限公司 | Newly added garbage identification method and device, computer equipment and storage medium |
| CN114077877B (en) * | 2022-01-19 | 2022-05-13 | 人民中科(北京)智能技术有限公司 | Newly-added garbage identification method and device, computer equipment and storage medium |
| CN116342895A (en) * | 2023-05-31 | 2023-06-27 | 浙江联运知慧科技有限公司 | Method and system for improving sorting efficiency of renewable resources based on AI (advanced technology attachment) processing |
| CN116342895B (en) * | 2023-05-31 | 2023-08-11 | 浙江联运知慧科技有限公司 | Method and system for improving sorting efficiency of renewable resources based on AI (advanced technology attachment) processing |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112651318A (en) | Image recognition-based garbage classification method, device and system | |
| Rabano et al. | Common garbage classification using mobilenet | |
| CN110490246B (en) | Garbage category determination method and device, storage medium and electronic equipment | |
| CN108706246A (en) | Intelligent refuse classification reclaimer, control method, device and storage medium | |
| WO2021012761A1 (en) | Garbage classification method, apparatus and device, and storage medium | |
| CN112298844B (en) | Garbage classification supervision method and device | |
| CN113003054A (en) | Garbage classification method | |
| CN113052005B (en) | Garbage sorting method and garbage sorting device for household service | |
| CN113003031B (en) | Garbage collection method, garbage collection robot and garbage collection system | |
| Thibuy et al. | Holistic solution design and implementation for smart city recycle waste management case study: Saensuk city | |
| Song et al. | Optimization of CNN-based garbage classification model | |
| CN113562357B (en) | Garbage recognition method | |
| CN117465862A (en) | Intelligent garbage can management method and device | |
| CN113743404A (en) | Intelligent garbage collection and transportation vehicle garbage image classification method | |
| CN112215147A (en) | Tracing method, device, equipment and storage medium for garbage throwing position | |
| CN112183385A (en) | Target object processing method and device | |
| Nurahmadan et al. | A mobile based waste classification using mobilenets-v1 architecture | |
| CN115393780A (en) | Garbage foreign matter identification method and device, electronic equipment and storage medium | |
| CN113378627A (en) | Automatic garbage classification method and device based on DSP chip and single chip microcomputer control | |
| CN113619950B (en) | Garbage classification putting intelligent supervision system and method based on deep learning | |
| CN114612730A (en) | Method and device for detecting household garbage classification generation proportion | |
| LU500437B1 (en) | System and method for intelligent garbage sorting based on machine vision | |
| CN113428525A (en) | Garbage classification system based on computer vision | |
| CN117181763B (en) | Renewable resource recycling inspection system based on machine vision recognition | |
| CN114202448A (en) | Garbage classification and recovery processing method, terminal, garbage can, server and medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |