CN111056191A - Intelligent garbage collection system capable of classifying and delivering multiple garbage - Google Patents
Intelligent garbage collection system capable of classifying and delivering multiple garbage Download PDFInfo
- Publication number
- CN111056191A CN111056191A CN202010036883.7A CN202010036883A CN111056191A CN 111056191 A CN111056191 A CN 111056191A CN 202010036883 A CN202010036883 A CN 202010036883A CN 111056191 A CN111056191 A CN 111056191A
- Authority
- CN
- China
- Prior art keywords
- garbage
- classification
- image
- size
- intelligent
- 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
Images
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F1/00—Refuse receptacles; Accessories therefor
- B65F1/14—Other constructional features; Accessories
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F1/00—Refuse receptacles; Accessories therefor
- B65F1/0033—Refuse receptacles; Accessories therefor specially adapted for segregated refuse collecting, e.g. receptacles with several compartments; Combination of receptacles
- B65F1/0053—Combination of several receptacles
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F1/00—Refuse receptacles; Accessories therefor
- B65F1/14—Other constructional features; Accessories
- B65F1/16—Lids or covers
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F1/00—Refuse receptacles; Accessories therefor
- B65F1/0033—Refuse receptacles; Accessories therefor specially adapted for segregated refuse collecting, e.g. receptacles with several compartments; Combination of receptacles
- B65F2001/008—Means for automatically selecting the receptacle in which refuse should be placed
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F2210/00—Equipment of refuse receptacles
- B65F2210/138—Identification means
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F2210/00—Equipment of refuse receptacles
- B65F2210/176—Sorting means
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B65—CONVEYING; PACKING; STORING; HANDLING THIN OR FILAMENTARY MATERIAL
- B65F—GATHERING OR REMOVAL OF DOMESTIC OR LIKE REFUSE
- B65F2210/00—Equipment of refuse receptacles
- B65F2210/184—Weighing means
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02W—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO WASTEWATER TREATMENT OR WASTE MANAGEMENT
- Y02W30/00—Technologies for solid waste management
- Y02W30/10—Waste collection, transportation, transfer or storage, e.g. segregated refuse collecting, electric or hybrid propulsion
Landscapes
- Engineering & Computer Science (AREA)
- Mechanical Engineering (AREA)
- Sorting Of Articles (AREA)
Abstract
The invention discloses an intelligent garbage collection system capable of classifying and delivering multiple garbage, which comprises a delivery port device, a base and a power supply device, wherein the base is positioned below the delivery port device; the delivery opening device comprises a cover body and a garbage input opening, a size classification roller is arranged below the garbage input opening, the size classification roller and the inner wall of the garbage input opening form a size control ring, the size classification roller is fixed through fixing devices at two ends, a motor is arranged on the fixing devices and used for driving the size classification roller, a garbage storage sieve and a plurality of garbage cans are arranged in the base, and a laser sensing device is arranged in the base; two claw mechanical arms are arranged above the garbage storage sieve, and a garbage image acquisition and object detection system is further arranged above the garbage storage sieve and used for classifying garbage. The system can classify and accurately deliver a plurality of different types of garbage or bagged garbage at the same time, manual participation is not needed, and the garbage can be thrown in from an entrance.
Description
Technical Field
The invention relates to an intelligent garbage collection system capable of classifying and delivering multiple garbage, and belongs to the technical field of intelligent public equipment.
Background
At present, when garbage is thrown, the classification garbage bins on both sides of a road generally judge which kind of garbage is thrown according to subjective judgment of people and throw the garbage into corresponding garbage bins, and the mode is applicable to single garbage or scattered garbage. However, when delivering the bagged garbage, most of the garbage in the bags is in a large quantity and in a large class, if the garbage bags are opened again for classified delivery of the garbage, the steps are complicated and are not applicable.
At present, domestic researchers combine the technique of intelligent manufacturing, have designed a great deal of intelligent classification's garbage bin according to actual demand. For example, the Chinese patent application No. 201910590099.8 discloses an intelligent classification garbage can and a classification method thereof, wherein after garbage is classified by an identification algorithm, a partition plate is overturned and poured, and the operation is simple. However, this method also has the disadvantage that if only one type of garbage is on the dumping partition, it is feasible, and if there are multiple types of garbage or the whole bag of garbage, the classification purpose cannot be achieved. For example, the Chinese patent application No. 201910883247.5 discloses an intelligent classification garbage can and an intelligent classification system, wherein a plurality of cameras are used for collecting images of garbage thrown into the can, the images are subjected to a series of image processing algorithms such as brightness enhancement, gray level correction, noise filtration and template matching, a turning plate is driven to rotate through a rotating motor, the garbage is conveyed to a U-shaped plate and is sequentially delivered to corresponding classification boxes for compression, the garbage classification process is complex, the cost is relatively high, and other small-size garbage is likely to be conveyed to the U-shaped plate together in the turning plate rotating process, so that the problem of inaccurate classification is caused.
With the development of the full convolution neural network technology based on deep learning, the accuracy of image recognition is higher and higher, even the recognition of some categories exceeds that of human beings, excessive preprocessing steps are not needed in image processing, and therefore the recognition efficiency can be improved, and the algorithm content can be simplified.
The existing intelligent classification garbage can is mainly used for treating single garbage or a plurality of bulk garbage, and the application range is narrow. For example, garbage delivered in offices or at home is generally discarded in bags. This is a challenge for intelligently sorting trash cans. Therefore, there is a need for a trash can with improved design, which combines the current object detection and intelligent robot technology to design a universally applicable intelligent classification trash can.
Disclosure of Invention
To above demand and not enough, designed one kind to single, scattered a plurality of or the all suitable intelligent classification garbage bin of rubbish of bagging-off, can classify and accurately deliver a plurality of rubbish of different classes or rubbish in bags simultaneously, need not artifical the participation, only need with rubbish from the entrance drop into can.
The technical scheme of the invention is as follows:
an intelligent garbage collection system capable of classifying and delivering multiple garbage comprises a delivery port device, a base positioned below the delivery port device and a power supply device for supplying power to the whole system;
the delivery port device comprises a cover body and a garbage input port arranged above the cover body, a size classification roller is arranged below the garbage input port, the size classification roller and the inner wall of the garbage input port form a size control ring, the size classification roller is fixed through fixing devices at two ends, and a motor is arranged on each fixing device and used for driving the size classification roller;
the garbage storage sieve is fixed on the side wall of the base and is provided with a laser sensing device, and when garbage falls off, the whole system starts to work;
two claw mechanical arms are arranged above the garbage storage sieve and used for grabbing garbage on the garbage storage sieve; a garbage image acquisition and object detection system is also arranged above the garbage storage sieve and is used for classifying garbage; the garbage image collecting and object detecting system is used for positioning and classifying collected garbage images through a neural network to obtain distribution positions and categories of garbage on a garbage storage sieve, positions corresponding to various types of garbage are transmitted to the microprocessor in an electric signal mode, the microprocessor transmits signals to the two-claw mechanical arm, and the two-claw mechanical arm grabs the garbage on the garbage storage sieve according to the positions and moves to the position above a garbage can with a corresponding number according to the categories to deliver the garbage.
Preferably, one of the two-jaw mechanical arm is an electromagnetic mechanical jaw, and the electromagnetic mechanical jaw can adsorb metal iron when being powered on and drop the metal iron when being powered off.
Preferably, the bottom of every garbage bin is provided with the device that vibrates, vibrate the device and include elastic support machine structure and the motor screw lead screw device that is used for reciprocating, when rubbish delivery finishes at every turn, stop at the garbage bin and the automatic trigger switch of running clearance vibrates the staving and makes the interior rubbish of bucket distribute evenly, with some little rubbish fills in the bucket, improve the space utilization degree.
The size of the garbage throwing opening is determined according to the actual use scene, and can be round or square, and the like, and the garbage throwing opening is judged according to the storage mode of the garbage in the garbage can in the area, such as the size of single garbage and the size of bagged garbage. If the garbage thrown in the area every day is small and most of the garbage is single garbage, the designed throwing opening is small in size, such as 500ml beverage bottles, pop cans and the like, and the garbage can is applied to scenes, such as garbage cans on two sides of a highway, in a campus or in a scenic spot; if the size of the thrown garbage is large and the garbage is partially bagged garbage, for example, a plastic bag for a circular paper basket with the diameter of 26cm or 32cm is used, and the garbage can is applied to scenes such as garbage cans in office buildings or downstairs of a community. The size of the garbage can feeding opening is set to be 1.5 times of the diameter of the fed garbage.
When the garbage throwing opening is circular, the size classification rollers are positioned between the size control rings, the size classification rollers and the inner wall of the circular garbage inlet form the size control rings, when the garbage is thrown into the barrel, firstly, the thrown garbage is primarily treated through the size classification rollers which are driven by the rotation of a motor, namely, single garbage or bagged garbage, and when the size of the garbage is smaller than the threshold value of the set size control rings, the garbage directly falls onto the garbage storage sieve; otherwise, the garbage bag is extruded by the size sorting roller to be broken, so that the garbage in the garbage bag is scattered and falls to the garbage storage sieve, and the position of the size sorting roller is 1/3 away from the center of the circle of the cylinder for throwing the garbage.
Preferably, the number of the garbage cans is four, and the garbage cans are used for containing recyclable garbage, harmful garbage, kitchen garbage and other garbage respectively.
The garbage categories collected by the four garbage cans are as follows:
and (3) garbage recycling: there are five major categories of waste paper, plastic, glass, metal and cloth.
Kitchen garbage: leftover, bone, leaf and peel of vegetable root.
Harmful garbage: comprises a battery, a fluorescent lamp tube, a bulb, a mercury thermometer, a paint bucket, partial household appliances, overdue medicines, overdue cosmetics and the like. Contains heavy metals harmful to human health, toxic substances or wastes causing real or potential harm to the environment.
And (3) other garbage: waste other than the above-mentioned types of waste.
Preferably, the power supply device is one or more of photovoltaic power supply, storage battery power supply and municipal power supply.
Preferably, the system further comprises a status detection system comprising a height signal sensor and an image status receptor; the height signal sensors are arranged at the 3/4 height of the wall of each garbage can, the height signal sensing system monitors the height of the garbage in the garbage can in real time, once the height reaches a limit value, the throwing port is closed, and meanwhile, a signal is sent to a nearby property management place so as to arrange special personnel to pour the garbage as soon as possible; the image state receptor regularly carries out self-checking on the collected images, if the phenomena that the resolution ratio of the images is too low or the images cannot be collected and the like are found, the input port is closed, and signals are fed back to a nearby property management position to automatically report and repair the images.
Preferably, the method for detecting and classifying the Garbage by the Garbage image collecting and object detecting system adopts a garpage Net algorithm, and comprises the following specific steps:
(1) establishing data sets of various garbage types; in order to train a full convolution neural network for object detection, a data set of recoverable garbage, harmful garbage, kitchen garbage and other garbage is established, the data set comprises 30K images of various garbage, wherein 8.2K images of the recoverable garbage, 7.8K images of the harmful garbage, 8K images of the kitchen garbage and the rest other garbage, and a weight model file obtained after the network is trained by using the images is provided for an object detection system 10 to realize the detection of various garbage stored on a screen;
(2) a camera of the garbage image acquisition and object detection system acquires a color image of garbage; inputting the obtained color image into a shallow convolutional neural network to extract some abstract semantic features of color pixels and pixel space positions, wherein the size of the image is kept unchanged in the convolutional neural network calculation process;
after each convolution layer, respectively using batch regularization and linear unit correction processing;
processing the convolutional layer after image input to obtain a 16-channel characteristic diagram of the original image, wherein the size of a sliding window is 0.2 times of the side length of the original image, the sliding window sequentially slides from left to right and from top to bottom, and the moving step length is 2;
(3) the sliding window detector with the size is used for processing the characteristic diagram of the collected color garbage image, and the two processes are divided into:
(3-1) converting the characteristic diagram of the color garbage image into an 8-bit gray level image, calculating a gray level co-occurrence matrix in a sliding window detector, and respectively calculating entropy representing pixel complexity in a sliding window, autocorrelation of texture consistency and an inverse difference matrix representing the local change speed of image texture so as to preliminarily determine the object type of garbage in the image;
(3-2) respectively calculating a directional gradient histogram and further performing convolution processing on the feature map in the window, fusing the features obtained by the directional gradient histogram and the further convolution processing, obtaining a classification result by using cross entropy as a classification function, and obtaining a bounding box of a final image by using logistic regression; the frames of the classification result may include a plurality of coincident frames, the coincident areas of different windows are calculated by adopting a non-maximum value inhibition method, and if the coincident areas are larger than a set threshold value, one of the surrounding frames is deleted; the operations are circulated until all the calculations are completed;
(3-3) comparing the classification result of (3-2) with the determined object classification of the step (3-1) for the classification contained in each picture, if the classification result is the same as the result, the classification is correct, otherwise, the classification result is wrong, if the classification result is wrong, the classification result returns to a sliding window, the window size is changed to be 0.5 times of the side length, and if the result obtained by the gray level co-occurrence matrix is not the same as the network classification result, the processing is carried out according to the network classification result;
(3-4) training the Garbage classification Garpage Net algorithm by adopting an error back propagation algorithm and taking a cross entropy function as a loss function again according to the steps, and iterating for 50K times or stopping training when the classification loss reaches about 0.05;
(4) after the training is finished, importing the weight file trained in the step (3-4) into a garbage classification system for use to obtain a classification result;
(5) and (5) grabbing and delivering the garbage cans to the corresponding garbage cans according to the classification result of the step (4) and the positions of the garbage cans and the surrounding frame.
The working process of the system of the invention is as follows:
bagged rubbish when dropping into the bucket, the size control ring of entrance below can be with sack extrusion deformation, the sack breaks, rubbish in the sack can drop in the sieve is deposited to below rubbish, rubbish is deposited the sieve and can be at first filtered moisture or tiny object residue in the rubbish, rubbish is deposited the rubbish image acquisition of sieve top and can flash light with object detection system's camera and shoot rubbish image, through the full convolution neural network of object detection based on degree of depth study, the rubbish classification in the image is discerned to the mode of frame, and fix a position, two claw arms this moment can initiatively snatch the rubbish of each classification and according to the reference numeral of discernment, drop into in the garbage bin that corresponds.
When the garbage bin was deposited to full, when the staff cleared up rubbish, can pull down the sieve of depositing and clear up, avoid the screen cloth to block up.
The invention achieves the following beneficial effects:
1. and various garbage data sets are established for training and verifying algorithms for garbage classification and positioning, so that the recognition precision is improved, the complexity of the algorithms is simplified, and the image preprocessing and the like in the early stage are not needed. Since the data set is important to the accuracy of the algorithm, the data set is enhanced, including rotation, contrast adjustment, brightness adjustment, cropping, and the like.
2. The garbage can process a plurality of types and a plurality of quantities of garbage simultaneously, is not only limited to single or a plurality of garbage, but also can be used for processing bagged garbage, has simple operation and superior cost, and does not need manual participation.
3. The size of the garbage throwing port and the size of the garbage storage box are designed according to a statistical method, so that the actual weight of various garbage throwing can be more approximate, the garbage box cleaning period is prolonged, and manpower and material resources are saved.
Drawings
FIG. 1 is a schematic view of the overall structure of the present invention;
FIG. 2 is a schematic illustration of FIG. 1 with the delivery port device removed;
FIG. 3 is a schematic view of the structure of a delivery port device;
FIG. 4 is a schematic structural view of a base;
fig. 5 is a flowchart of a garpage _ Net algorithm for Garbage classification.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
As shown in fig. 1 and 2, an intelligent garbage collection system capable of multi-garbage classification and delivery comprises a delivery opening device, a base positioned below the delivery opening device and a power supply device for supplying power to the whole system;
as shown in fig. 3, the delivery opening device comprises a cover body and a garbage input opening 12 arranged above the cover body, a size classification roller 7 is arranged below the garbage input opening 12, the size classification roller 7 and the inner wall of the garbage input opening 12 form a size control ring 9, the size classification roller 7 is fixed by a fixer 8 at two ends, and a motor 6 is arranged on the fixer 8 for driving the size classification roller 7;
as shown in fig. 4, a garbage storage sieve 4 and a plurality of garbage cans 3 arranged at the lower end of the garbage storage sieve 4 are arranged in the base, the garbage storage sieve 4 is positioned at the lower end of the garbage putting-in opening 12, the garbage storage sieve 12 is fixed on the side wall of the base and is provided with a laser sensing device, and when garbage falls, the whole system starts to work;
a two-claw mechanical arm 5 is arranged above the garbage storage sieve 4 and used for grabbing garbage on the garbage storage sieve 4; a garbage image acquisition and object detection system 10 is also arranged above the garbage storage sieve 4 and is used for classifying garbage; the garbage image collecting and object detecting system 10 positions and classifies collected garbage images through a neural network to obtain distribution positions and types of garbage on the garbage storage sieve 4, positions corresponding to various types of garbage are transmitted to the microprocessor 11 in an electric signal mode, the microprocessor 11 transmits signals to the two-claw mechanical arm 5, and the two-claw mechanical arm 5 grabs the garbage on the garbage storage sieve 4 according to the positions and moves to the position above the garbage can 3 with the corresponding number according to the types to deliver the garbage.
Preferably, one of the two-jaw mechanical arm 5 is an electromagnetic mechanical jaw, and can adsorb metal iron when the two-jaw mechanical arm is powered on, and the metal iron falls off when the two-jaw mechanical arm is powered off.
Preferably, the bottom of every garbage bin 3 is provided with the device that vibrates, vibrate the device and include elastic support machine structure 1 and be used for the motor screw lead screw device 2 that reciprocates, when rubbish delivery finishes at every turn, stop at the garbage bin at the working gap automatic triggering switch and vibrate the staving and make the interior rubbish of bucket distribute evenly, fill some little rubbish in the bucket, improve the space utilization.
The garbage throwing port 12 is used for determining the size of the throwing port according to the actual use scene, can be in the shape of a circle or a square and the like, and is judged according to the storage mode of the garbage in the garbage can in the area, such as the size of single garbage and the size of bagged garbage. If the garbage thrown in the area every day is small and most of the garbage is single garbage, the designed throwing opening is small in size, such as 500ml beverage bottles, pop cans and the like, and the garbage can is applied to scenes, such as garbage cans on two sides of a highway, in a campus or in a scenic spot; if the size of the thrown garbage is large and the garbage is partially bagged garbage, for example, a plastic bag for a circular paper basket with the diameter of 26cm or 32cm is used, and the garbage can is applied to scenes such as garbage cans in office buildings or downstairs of a community. The size of the garbage can feeding opening is set to be 1.5 times of the diameter of the fed garbage.
When the garbage throwing-in opening 12 is circular, the size classifying rollers 7 are positioned between the size control rings 9, the size control rings are formed by the size classifying rollers 7 and the inner wall of the circular garbage throwing-in opening 12, when the garbage is thrown into the barrel, firstly, the size classifying rollers 7 pass through the size classifying rollers 7 and are driven by the rotation of the motor 6 to carry out primary treatment on the thrown-in garbage, namely single garbage or bagged garbage, and when the size of the garbage is smaller than the threshold value of the set size control ring, the garbage directly falls onto a garbage storage sieve; otherwise, the garbage bag is extruded by the size classifying roller 7 to be broken, so that the garbage in the garbage bag is scattered and falls to the garbage storage sieve 4, and the position of the size classifying roller 7 is away from 1/3 of the center of the cylinder for throwing the garbage.
Preferably, the number of the garbage cans 3 is four, and the garbage cans are respectively used for containing recyclable garbage, harmful garbage, kitchen garbage and other garbage.
The garbage categories collected by the four garbage cans are as follows:
and (3) garbage recycling: there are five major categories of waste paper, plastic, glass, metal and cloth.
Kitchen garbage: leftover, bone, leaf and peel of vegetable root.
Harmful garbage: comprises a battery, a fluorescent lamp tube, a bulb, a mercury thermometer, a paint bucket, partial household appliances, overdue medicines, overdue cosmetics and the like. Contains heavy metals harmful to human health, toxic substances or wastes causing real or potential harm to the environment.
And (3) other garbage: waste other than the above-mentioned types of waste.
Preferably, the power supply device is one or more of photovoltaic power supply, storage battery power supply and municipal power supply.
Preferably, the system further comprises a status detection system comprising a height signal sensor and an image status receptor; the height signal sensors are arranged at the 3/4 height of the wall of each garbage can, the height signal sensing system monitors the height of the garbage in the garbage can in real time, once the height reaches a limit value, the throwing port is closed, and meanwhile, a signal is sent to a nearby property management place so as to arrange special personnel to pour the garbage as soon as possible; the image state receptor regularly carries out self-checking on the collected images, if the phenomena that the resolution ratio of the images is too low or the images cannot be collected and the like are found, the input port is closed, and signals are fed back to a nearby property management position to automatically report and repair the images.
As shown in fig. 5, preferably, the method for detecting and classifying trash by the trash image collecting and object detecting system 10 adopts a garpage Net algorithm, and includes the following specific steps:
(1) establishing data sets of various garbage types;
(2) a camera of the garbage image acquisition and object detection system acquires a color image of garbage; inputting the obtained color image into a shallow convolutional neural network to extract some abstract semantic features of color pixels and pixel space positions, wherein the size of the image is kept unchanged in the convolutional neural network calculation process;
after each convolution layer, respectively using batch regularization and linear unit correction processing;
processing the convolutional layer after image input to obtain a 16-channel characteristic diagram of the original image, wherein the size of a sliding window is 0.2 times of the side length of the original image, the sliding window sequentially slides from left to right and from top to bottom, and the moving step length is 2;
(3) the sliding window detector with the size is used for processing the characteristic diagram of the collected color garbage image, and the two processes are divided into:
(3-1) converting the characteristic diagram of the color garbage image into an 8-bit gray level image, calculating a gray level co-occurrence matrix in a sliding window detector, and respectively calculating entropy representing pixel complexity in a sliding window, autocorrelation of texture consistency and an inverse difference matrix representing the local change speed of image texture so as to preliminarily determine the object type of garbage in the image;
(3-2) respectively calculating a directional gradient histogram and further performing convolution processing on the feature map in the window, fusing the features obtained by the directional gradient histogram and the further convolution processing, obtaining a classification result by using cross entropy as a classification function, and obtaining a bounding box of a final image by using logistic regression; the frames of the classification result may include a plurality of coincident frames, the coincident areas of different windows are calculated by adopting a non-maximum value inhibition method, and if the coincident areas are larger than a set threshold value, one of the surrounding frames is deleted; the operations are circulated until all the calculations are completed;
(3-3) comparing the classification result of (3-2) with the determined object classification of the step (3-1) for the classification contained in each picture, if the classification result is the same as the result, the classification is correct, otherwise, the classification result is wrong, if the classification result is wrong, the classification result returns to a sliding window, the window size is changed to be 0.5 times of the side length, and if the result obtained by the gray level co-occurrence matrix is not the same as the network classification result, the processing is carried out according to the network classification result;
(3-4) training the Garbage classification Garpage Net algorithm by adopting an error back propagation algorithm and taking a cross entropy function as a loss function again according to the steps, and iterating for 50K times or stopping training when the classification loss reaches about 0.05;
(4) after the training is finished, importing the weight file trained in the step (3-4) into a garbage classification system for use to obtain a classification result;
(5) and (5) grabbing and delivering the garbage cans to the corresponding garbage cans according to the classification result of the step (4) and the positions of the garbage cans and the surrounding frame.
The working process of the system of the invention is as follows:
bagged rubbish when dropping into the bucket, the size control ring of entrance below can be with sack extrusion deformation, the sack breaks, rubbish in the sack can drop in the sieve is deposited to below rubbish, rubbish is deposited the sieve and can be at first filtered moisture or tiny object residue in the rubbish, rubbish is deposited the rubbish image acquisition of sieve top and can flash light with object detection system's camera and shoot rubbish image, through the full convolution neural network of object detection based on degree of depth study, the rubbish classification in the image is discerned to the mode of frame, and fix a position, two claw arms this moment can initiatively snatch the rubbish of each classification and according to the reference numeral of discernment, drop into in the garbage bin that corresponds.
When the garbage bin was deposited to full, when the staff cleared up rubbish, can pull down the sieve of depositing and clear up, avoid the screen cloth to block up.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (8)
1. The utility model provides an intelligent garbage collection system that can many garbage classification and deliver which characterized in that: the system comprises a delivery port device, a base positioned below the delivery port device and a power supply device for supplying power to the whole system;
the delivery port device comprises a cover body and a garbage input port arranged above the cover body, a size classification roller is arranged below the garbage input port, the size classification roller and the inner wall of the garbage input port form a size control ring, the size classification roller is fixed through fixing devices at two ends, and a motor is arranged on each fixing device and used for driving the size classification roller;
the garbage storage sieve is fixed on the side wall of the base and is provided with a laser sensing device, and when garbage falls off, the whole system starts to work;
two claw mechanical arms are arranged above the garbage storage sieve and used for grabbing garbage on the garbage storage sieve; a garbage image acquisition and object detection system is also arranged above the garbage storage sieve and is used for classifying garbage; the garbage image collecting and object detecting system is used for positioning and classifying collected garbage images through a neural network to obtain distribution positions and categories of garbage on a garbage storage sieve, positions corresponding to various types of garbage are transmitted to the microprocessor in an electric signal mode, the microprocessor transmits signals to the two-claw mechanical arm, and the two-claw mechanical arm grabs the garbage on the garbage storage sieve according to the positions and moves to the position above a garbage can with a corresponding number according to the categories to deliver the garbage.
2. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: one of the two claw arms is an electromagnetic mechanical claw, and can adsorb metal iron when the two claw arms are powered on and drop the metal iron when the two claw arms are powered off.
3. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: the bottom of every garbage bin is provided with the device that vibrates, vibrate the device and include elastic support machine structure and the motor screw thread lead screw device that is used for reciprocating, when rubbish delivery finishes at every turn, stop at the garbage bin working gap automatic trigger switch and vibrate the staving and make the interior rubbish of bucket distribute evenly, fill some little rubbish in the bucket, improve the space utilization.
4. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: the garbage throwing opening is circular, and the position of the size classification roller is 1/3 away from the center of the garbage throwing opening.
5. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: the garbage can is four in number and is used for containing recoverable garbage, harmful garbage, kitchen garbage and other garbage respectively.
6. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: the power supply device is one or a combination of photovoltaic power supply, storage battery power supply and municipal power supply.
7. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: the system further comprises a status detection system comprising a height signal sensor and an image status receptor; the height signal sensors are arranged at the 3/4 height of the wall of each garbage can, the height signal sensing system monitors the height of the garbage in the garbage can in real time, once the height reaches a limit value, the throwing port is closed, and meanwhile, a signal is sent to a nearby property management place so as to arrange special personnel to pour the garbage as soon as possible; the image state receptor regularly carries out self-checking on the collected images, if the phenomena that the resolution ratio of the images is too low or the images cannot be collected and the like are found, the input port is closed, and signals are fed back to a nearby property management position to automatically report and repair the images.
8. The intelligent garbage collection system capable of multi-garbage classification and delivery according to claim 1, characterized in that: the method for detecting and classifying the Garbage by the Garbage image acquisition and object detection system adopts a Garpage Net algorithm, and comprises the following specific steps:
(1) establishing data sets of various garbage types;
(2) a camera of the garbage image acquisition and object detection system acquires a color image of garbage; inputting the obtained color image into a shallow convolutional neural network to extract some abstract semantic features of color pixels and pixel space positions, wherein the size of the image is kept unchanged in the convolutional neural network calculation process;
after each convolution layer, respectively using batch regularization and linear unit correction processing;
processing the convolutional layer after image input to obtain a 16-channel characteristic diagram of the original image, wherein the size of a sliding window is 0.2 times of the side length of the original image, the sliding window sequentially slides from left to right and from top to bottom, and the moving step length is 2;
(3) the sliding window detector with the size is used for processing the characteristic diagram of the collected color garbage image, and the two processes are divided into:
(3-1) converting the characteristic diagram of the color garbage image into an 8-bit gray level image, calculating a gray level co-occurrence matrix in a sliding window detector, and respectively calculating entropy representing pixel complexity in a sliding window, autocorrelation of texture consistency and an inverse difference matrix representing the local change speed of image texture so as to preliminarily determine the object type of garbage in the image;
(3-2) respectively calculating a directional gradient histogram and further performing convolution processing on the feature map in the window, fusing the features obtained by the directional gradient histogram and the further convolution processing, obtaining a classification result by using cross entropy as a classification function, and obtaining a bounding box of a final image by using logistic regression; the frames of the classification result may include a plurality of coincident frames, the coincident areas of different windows are calculated by adopting a non-maximum value inhibition method, and if the coincident areas are larger than a set threshold value, one of the surrounding frames is deleted; the operations are circulated until all the calculations are completed;
(3-3) comparing the classification result of (3-2) with the determined object classification of the step (3-1) for the classification contained in each picture, if the classification result is the same as the result, the classification is correct, otherwise, the classification result is wrong, if the classification result is wrong, the classification result returns to a sliding window, the window size is changed to be 0.5 times of the side length, and if the result obtained by the gray level co-occurrence matrix is not the same as the network classification result, the processing is carried out according to the network classification result;
(3-4) training the Garbage classification Garpage Net algorithm by adopting an error back propagation algorithm and taking a cross entropy function as a loss function again according to the steps, and iterating for 50K times or stopping training when the classification loss reaches about 0.05;
(4) after the training is finished, importing the weight file trained in the step (3-4) into a garbage classification system for use to obtain a classification result;
(5) and (5) grabbing and delivering the garbage cans to the corresponding garbage cans according to the classification result of the step (4) and the positions of the garbage cans and the surrounding frame.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010036883.7A CN111056191A (en) | 2020-01-14 | 2020-01-14 | Intelligent garbage collection system capable of classifying and delivering multiple garbage |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010036883.7A CN111056191A (en) | 2020-01-14 | 2020-01-14 | Intelligent garbage collection system capable of classifying and delivering multiple garbage |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111056191A true CN111056191A (en) | 2020-04-24 |
Family
ID=70307333
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010036883.7A Pending CN111056191A (en) | 2020-01-14 | 2020-01-14 | Intelligent garbage collection system capable of classifying and delivering multiple garbage |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111056191A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112093307A (en) * | 2020-08-07 | 2020-12-18 | 张国靖 | Based on internet + intelligent garbage bin |
CN112407680A (en) * | 2020-11-06 | 2021-02-26 | 钱继红 | Intensive intelligent garbage collection equipment and working method thereof |
CN112764378A (en) * | 2021-01-04 | 2021-05-07 | 国泰百安信息技术有限公司 | Intelligent community management system based on Internet of things |
CN113173349A (en) * | 2021-05-06 | 2021-07-27 | 上海市格致中学 | Garbage classification system and method |
CN113306909A (en) * | 2021-05-31 | 2021-08-27 | 江苏奥帆科技有限公司 | Intelligent garbage classification and recovery control system and method |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004217420A (en) * | 2003-01-16 | 2004-08-05 | Masaru Sakai | Separating type garbage box using vinyl bag of supermarket |
CN205686939U (en) * | 2016-06-26 | 2016-11-16 | 济宁市瑞驰环保科技有限公司 | A kind of intelligent classification Environmental-protection garbage collection device |
CN109625680A (en) * | 2019-01-28 | 2019-04-16 | 中国地质大学(武汉) | A kind of intelligent recognition classification recycling dustbin |
CN109867059A (en) * | 2019-03-29 | 2019-06-11 | 克拉玛依市诚谊科技服务有限公司 | Garbage classification intelligence jettison system |
CN110155556A (en) * | 2019-04-30 | 2019-08-23 | 上海师范大学 | A kind of intelligent classification dustbin |
CN209796465U (en) * | 2019-04-29 | 2019-12-17 | 威海职业学院 | Biochemical medical waste material holds device |
CN110626662A (en) * | 2019-10-12 | 2019-12-31 | 张颢宸 | Image recognition-based garbage self-classification method and device |
-
2020
- 2020-01-14 CN CN202010036883.7A patent/CN111056191A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004217420A (en) * | 2003-01-16 | 2004-08-05 | Masaru Sakai | Separating type garbage box using vinyl bag of supermarket |
CN205686939U (en) * | 2016-06-26 | 2016-11-16 | 济宁市瑞驰环保科技有限公司 | A kind of intelligent classification Environmental-protection garbage collection device |
CN109625680A (en) * | 2019-01-28 | 2019-04-16 | 中国地质大学(武汉) | A kind of intelligent recognition classification recycling dustbin |
CN109867059A (en) * | 2019-03-29 | 2019-06-11 | 克拉玛依市诚谊科技服务有限公司 | Garbage classification intelligence jettison system |
CN209796465U (en) * | 2019-04-29 | 2019-12-17 | 威海职业学院 | Biochemical medical waste material holds device |
CN110155556A (en) * | 2019-04-30 | 2019-08-23 | 上海师范大学 | A kind of intelligent classification dustbin |
CN110626662A (en) * | 2019-10-12 | 2019-12-31 | 张颢宸 | Image recognition-based garbage self-classification method and device |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112093307A (en) * | 2020-08-07 | 2020-12-18 | 张国靖 | Based on internet + intelligent garbage bin |
CN112407680A (en) * | 2020-11-06 | 2021-02-26 | 钱继红 | Intensive intelligent garbage collection equipment and working method thereof |
CN112764378A (en) * | 2021-01-04 | 2021-05-07 | 国泰百安信息技术有限公司 | Intelligent community management system based on Internet of things |
CN112764378B (en) * | 2021-01-04 | 2021-08-10 | 国泰百安信息技术有限公司 | Intelligent community management system based on Internet of things |
CN113173349A (en) * | 2021-05-06 | 2021-07-27 | 上海市格致中学 | Garbage classification system and method |
CN113306909A (en) * | 2021-05-31 | 2021-08-27 | 江苏奥帆科技有限公司 | Intelligent garbage classification and recovery control system and method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111056191A (en) | Intelligent garbage collection system capable of classifying and delivering multiple garbage | |
CN110053903B (en) | Household garbage classification and recovery management system and method | |
CN110697273B (en) | Household garbage intelligent recognition and automatic classification system and method based on iterative learning control | |
CN109201514B (en) | Waste sorting recycle method, garbage classification device and classified-refuse recovery system | |
CN109684979B (en) | Image recognition technology-based garbage classification method and device and electronic equipment | |
CN109606991B (en) | Intelligent garbage can and garbage classification method based on deep learning | |
CN214191039U (en) | Apparatus for sorting and collecting sorted waste | |
CN107054936A (en) | A kind of refuse classification prompting dustbin and system based on image recognition | |
CN112024424B (en) | Man-machine cooperation type garbage sorting system | |
CN111439500B (en) | Automatic garbage classification method and automatic garbage classification device | |
CA3079983A1 (en) | Systems and methods for detecting waste receptacles using convolutional neural networks | |
CN111559588A (en) | Intelligent garbage can for classified garbage throwing and classified garbage throwing method | |
CN113135368B (en) | Intelligent garbage front-end classification system and method | |
CN109261539A (en) | A kind of garbage sorting system and method for view-based access control model identification and convolutional neural networks | |
CN112830129A (en) | Scenic spot rubbish intelligent classification system | |
CN105858011A (en) | Garbage bin robot | |
CN211365766U (en) | Intelligent garbage classification and collection equipment and system | |
CN111583508A (en) | Garbage classification management system | |
CN110817187A (en) | Intelligent garbage classification and collection equipment and system | |
CN111753844A (en) | Dry and wet garbage classification method, classification box and classification system | |
CN113371363A (en) | Classified garbage can, intelligent classified garbage can based on deep learning and classification method | |
CN115724083A (en) | Garbage classification method, intelligent garbage can, storage medium and electronic device | |
CN217576626U (en) | Classification garbage can | |
CN108328170A (en) | Multifunctional intellectual classification environmental-protection box | |
CN211086960U (en) | Garbage recycling system of integrated intelligent street lamp |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200424 |