CN217576626U - Classification garbage can - Google Patents
Classification garbage can Download PDFInfo
- Publication number
- CN217576626U CN217576626U CN202121481089.XU CN202121481089U CN217576626U CN 217576626 U CN217576626 U CN 217576626U CN 202121481089 U CN202121481089 U CN 202121481089U CN 217576626 U CN217576626 U CN 217576626U
- Authority
- CN
- China
- Prior art keywords
- garbage
- conveyor belt
- classification
- basket
- primary conveyor
- 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.)
- Active
Links
Images
Classifications
-
- 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
- Refuse Collection And Transfer (AREA)
Abstract
The utility model relates to a classification garbage can, which realizes that three types of garbage can be automatically thrown into different garbage baskets according to classification by means of two-stage conveyor belts, and realizes the compression of recoverable garbage; in one case, manual classification and automatic framing can be realized by using a control button; in the other case, the automatic classification is carried out by utilizing a deep learning mode of a convolutional neural network based on machine vision, and the automatic basket entering is carried out. Compared with the prior art, the device of the utility model can conveniently realize manual or automatic garbage classification, and the garbage can be automatically put into a basket; particularly, in the case of automatic waste classification, it is possible to solve the problem of waste classification from the source without worrying about the problem of wrong waste classification, and it is possible to efficiently recover recyclable waste, particularly plastic bottles.
Description
Technical Field
The utility model belongs to the technical field of waste classification, a categorised garbage bin is related to, especially, relate to a categorised garbage bin and intelligent classification garbage bin based on degree of depth study.
Background
Along with the progress of society, the total amount of garbage is continuously increased every year, and the state also puts forward a garbage classification policy. But at present, garbage is more and more abundant, and the garbage is more and more difficult to classify, so that the subsequent garbage treatment is difficult. Although the garbage can be classified everywhere in a community and a street, the garbage classification standard is relatively disordered due to the accelerated pace of life at present, people hurry about throwing garbage and do not pay attention to which type the garbage is to be classified to directly throw the garbage into the garbage can.
At present, some devices capable of automatically classifying garbage also exist, but the garbage cannot be accurately sorted out.
SUMMERY OF THE UTILITY MODEL
The device of the utility model can conveniently classify garbage. Furthermore, the utility model discloses a device can utilize the visual identification principle, utilizes transport mechanism to carry corresponding garbage bin with rubbish. This device has still designed a compressing mechanism, carries out compression treatment to the beverage bottle, increases space utilization.
The utility model aims at providing a categorised garbage bin and intelligent classification garbage bin based on degree of depth study exactly.
The purpose of the utility model can be realized through the following technical scheme:
the utility model provides a classification garbage can, which comprises a can body, a garbage classification unit and a garbage classification and collection unit, wherein the garbage classification unit and the garbage classification and collection unit are arranged inside the can body one above the other;
the top of the barrel body is provided with a garbage inlet,
the garbage classification unit comprises a primary conveyor belt, a compression mechanism positioned below the first end of the primary conveyor belt and a secondary conveyor belt positioned below the second end of the primary conveyor belt, the primary conveyor belt is positioned below the garbage inlet, the conveying directions of the primary conveyor belt and the secondary conveyor belt are arranged in a linear staggered manner,
the garbage classification collection unit comprises a recyclable garbage basket arranged below the compression mechanism, and a dry garbage basket and a wet garbage basket which are respectively arranged below the first end and the second end of the secondary conveyor belt.
Preferably, the garbage classification unit further comprises a control unit, the control unit comprises a controller and a garbage classification button assembly, the controller is electrically connected with the first-level conveyor belt, the second-level conveyor belt, the compression mechanism and the garbage classification button assembly respectively, and the garbage classification button assembly is composed of a wet garbage button, a dry garbage button and a recyclable garbage button and is used for controlling garbage entering the garbage inlet to enter a corresponding garbage basket in the garbage classification collection unit.
Preferably, the straight lines of the conveying directions of the primary conveyor belt and the secondary conveyor belt are perpendicular to each other in the horizontal plane.
Preferably, the primary conveyor belt and the secondary conveyor belt are both provided with belts and driving mechanisms for driving the belts to convey forwards and backwards, and the driving mechanisms comprise driven wheels, driving wheels and driving motors for driving the driving wheels to rotate forwards and backwards.
Preferably, the compression mechanism is a pop can compression mechanism.
Preferably, the garbage inlet is provided with an openable box cover.
Compared with the prior art, through the above technical scheme of the utility model, the user when using categorised garbage bin, only need from a rubbish entry throw away rubbish to the garbage bin can, then can be according to the type of rubbish, only need press the button once, can realize that rubbish is automatic to get into in the corresponding rubbish basket. The garbage can has a more compact structure, is easier to arrange and is more convenient to use. And all the garbage baskets are arranged in a closed garbage can, so that the garbage can is more attractive. The problem of set up that three independent garbage bins occupy a large space, arrange inconveniently is avoided.
The utility model provides an intelligent classification garbage can based on deep learning, which comprises a can body, a garbage classification unit and a garbage classification and collection unit, wherein the garbage classification unit and the garbage classification and collection unit are positioned inside the can body one above the other;
the top of the barrel body is provided with a garbage inlet,
the garbage classification unit comprises a primary conveyor belt, a compression mechanism positioned below the first end of the primary conveyor belt, a secondary conveyor belt positioned below the second end of the primary conveyor belt and an image acquisition camera positioned above the primary conveyor belt, wherein the primary conveyor belt is positioned below a garbage inlet, the conveying directions of the primary conveyor belt and the secondary conveyor belt are arranged in a linear staggered manner,
the garbage classification collection unit comprises a recyclable garbage basket arranged below the compression mechanism, and a dry garbage basket and a wet garbage basket which are respectively arranged below the first end and the second end of the secondary conveyor belt.
Preferably, the garbage classification unit further comprises a controller, and the controller is electrically connected with the first-level conveyor belt, the second-level conveyor belt, the compression mechanism and the image acquisition camera respectively.
Preferably, the image acquisition camera is used for acquiring image information of garbage falling onto the primary conveyor belt, and the controller is used for receiving the image information of the garbage, identifying the image information of the garbage based on deep learning, judging the garbage type, and further controlling the primary conveyor belt, the secondary conveyor belt and/or the compression mechanism to work so that the garbage enters the corresponding garbage basket.
Preferably, the straight lines of the conveying directions of the primary conveyor belt and the secondary conveyor belt are perpendicular to each other in the horizontal plane.
Preferably, the primary conveyor belt and the secondary conveyor belt are both provided with belts and driving mechanisms used for driving the belts to convey in the forward and reverse directions, and the driving mechanisms comprise driven wheels, driving wheels and driving motors used for driving the driving wheels to rotate in the forward and reverse directions.
Preferably, the compression mechanism is a pop can compression mechanism.
Preferably, the garbage inlet is provided with an openable box cover.
The utility model discloses the third aspect provides a classification method of intelligent classification garbage bin based on degree of depth study, including following step:
s1: the garbage falls onto the primary conveyor belt through the garbage inlet;
s2: the image acquisition camera samples, gathers the image information of the rubbish that falls on the one-level conveyer belt, and the controller receives the image information of rubbish to contrast in with the storehouse, judge the rubbish type:
if the garbage is recyclable garbage, the primary conveyor belt drives the garbage to convey towards the first end of the primary conveyor belt, the garbage is conveyed into the compressing mechanism, the compressing mechanism compresses the garbage, the garbage falls into a recyclable garbage basket after the garbage is completely processed,
if the garbage is dry garbage, the primary conveyor belt drives the garbage to be conveyed towards the second end of the primary conveyor belt, the garbage is conveyed onto the secondary conveyor belt, the secondary conveyor belt drives the garbage to be conveyed towards the first end of the secondary conveyor belt, the garbage is conveyed into a dry garbage basket,
if the garbage is wet garbage, the primary conveyor belt drives the garbage to be conveyed towards the second end of the primary conveyor belt, the garbage is conveyed onto the secondary conveyor belt, the secondary conveyor belt drives the garbage to be conveyed towards the second end of the secondary conveyor belt, and the garbage is conveyed into a wet garbage basket;
and the step S2 is carried out based on machine vision and by utilizing a deep learning mode of a convolutional neural network.
Compared with the prior art, through adopting the above technical scheme of the utility model, have a rain beneficial effect:
1. the garbage classification problem is solved from the source: although a lot of garbage is collected in public places and the garbage is complicated, the garbage at home has the same problem. The utility model discloses can use waste classification's idea in the family of resident household, solve waste classification's problem from the source, provide convenience for later waste classification handles.
2. The problem of garbage classification errors does not need to be worried about: another obstacle to garbage classification is that the lack of knowledge in garbage classification causes everyone to know which category we' garbage should belong to, and the work of this group just solves a problem effectively. We only need throw rubbish into the garbage bin, and the garbage bin can be put into different sub-garbage bins according to classification rule with rubbish classification automatically, need not everybody to think the classification problem again.
3. Plastic bottles were individually sorted for additional processing: as special recyclable garbage, the beverage bottle is independently placed, so that the garbage is convenient to recycle and is convenient for a waster to pick up.
4. The occupied space is improved to a certain extent, and the area of the space is reduced.
Drawings
Fig. 1 is a schematic structural view of a classification trash can according to embodiment 1 of the present invention.
Fig. 2 is the utility model discloses embodiment 2 is based on the structural schematic diagram of intelligent classification garbage bin of degree of depth study.
Fig. 3 is a schematic view of the top view structure of the garbage classification and collection unit of the present invention.
Fig. 4 is the structure diagram of the first-level conveyer belt and the second-level conveyer belt under different viewing angles of the present invention.
Fig. 5 is a schematic diagram of the classification method according to embodiment 2 of the present invention.
Fig. 6 is a screenshot of an example of model analysis in the model training process in embodiment 2 of the present invention.
Fig. 7, fig. 8, and fig. 9 are respectively an example screenshot of a training start state, a training middle state, and a training end state in a model training process in embodiment 2 of the present invention.
In the drawing, 1 is a can body, 11 is a garbage inlet, 2 is a garbage classification unit, 21 is a primary conveyor belt, 22 is a secondary conveyor belt, 23 is a compression mechanism, 201 is a belt, 202 is a driving wheel, 203 is a driven wheel, 204 is a driving motor, 3 is a garbage classification collection unit, 31 is a recyclable garbage basket, 32 is a dry garbage basket, 33 is a wet garbage basket, 4 is a controller, 5 is a garbage classification button assembly, and 6 is an image capture camera.
Detailed Description
The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.
Example 1
A classified garbage can is shown in figures 1 and 3 and comprises a can body 1, a garbage classification unit 2 and a garbage classification collection unit 3, wherein the garbage classification unit 2 and the garbage classification collection unit are arranged inside the can body 1 from top to bottom; the top of the barrel body 1 is provided with a garbage inlet 11, the garbage classification unit 2 comprises a first-level conveyor belt 21, a compression mechanism 23 and a second-level conveyor belt 22, the compression mechanism 23 is located below a first end of the first-level conveyor belt 21, the second-level conveyor belt 22 is located below a second end of the first-level conveyor belt 21, the first-level conveyor belt 21 is located below the garbage inlet 11, the conveying directions of the first-level conveyor belt 21 and the second-level conveyor belt 22 are arranged in a linear staggered mode, and the garbage classification collection unit 3 comprises a recyclable garbage basket 31, a dry garbage basket 32 and a wet garbage basket 33, the recyclable garbage basket 31 is arranged below the compression mechanism 23, and the dry garbage basket 32 and the wet garbage basket 33 are respectively arranged below a first end and a second end of the second-level conveyor belt 22.
More specifically, in the present embodiment:
the garbage classification unit preferably further comprises a control unit, the control unit comprises a controller 4 and a garbage classification button assembly 5, the controller 4 is electrically connected with the primary conveyor belt 21, the secondary conveyor belt 22, the compression mechanism 23 and the garbage classification button assembly 5 respectively, and the garbage classification button assembly 5 is composed of a wet garbage button, a dry garbage button and a recyclable garbage button and is used for controlling garbage entering the garbage inlet 11 to enter a corresponding garbage basket in the garbage classification collection unit 3. Controller 4 can adopt commercially available PLC controller or microcomputer controller, and controller 4's model can not be right the utility model discloses technical scheme's implementation causes the puzzlement.
When the user uses the classification garbage can, only the garbage needs to be thrown into the garbage can from one garbage inlet 11, and then the garbage can automatically enter the corresponding garbage basket by pressing the button according to the type of the garbage. The garbage can has a more compact structure, is easier to arrange and is more convenient to use. And all the garbage baskets are arranged in a closed garbage can, so that the garbage can is more attractive. The problems that three independent garbage cans occupy large space and are inconvenient to arrange are avoided.
An openable lid may also be provided at the waste inlet 11. In actual use, can also set up human sensor to be connected with controller 4, can realize that the automation of case lid is opened when sensing the user and being close to rubbish entry 11, the person of facilitating the use uses this garbage bin more.
The straight lines in which the conveying directions of the primary conveyor belt 21 and the secondary conveyor belt 22 are located are preferably perpendicular to each other in the horizontal plane. As shown in (a) to (d) of fig. 4. More preferably, the primary conveyor belt 21 and the secondary conveyor belt 22 both have a belt 201 and a driving mechanism for driving the belt 201 to convey forward and backward, and the driving mechanism is composed of a driven wheel 203, a driving wheel 202 and a driving motor 204 for driving the driving wheel 202 to rotate forward and backward.
The compression mechanism 23 is preferably a can compression mechanism. The compression mechanism of the pop can select related products in the prior art. Or the zip-top can compressing mechanism is provided with a cylinder body with an upper opening and a lower opening, the bottom of the cylinder body is provided with a bottom baffle plate which can be turned downwards, the inner side wall of the cylinder body is provided with a turnover compressing plate, the turnover compressing plate is driven by a cylinder arranged on the cylinder body in a penetrating way to be matched with the bottom baffle plate, garbage is compressed after falling into the cylinder body, and after the compression is finished, the bottom baffle plate is opened, so that the garbage falls into the recyclable garbage basket.
Putting harmful garbage: when harmful garbage is thrown in, the garbage should be taken and put lightly; putting the waste medicines together with packages; harmful garbage such as pressure tanks is thrown after the contents in the pressure tanks are emptied; when harmful garbage is generated in public places, the garbage is carried to a harmful garbage throwing point to be properly thrown; and for harmful garbage which is easy to damage, the garbage needs to be completely wrapped and then thrown.
Example 2
An intelligent classification garbage can based on deep learning is shown in fig. 2 and 3 and comprises a can body 1, a garbage classification unit 2 and a garbage classification collection unit 3, wherein the garbage classification unit 2 and the garbage classification collection unit are arranged inside the can body 1 one above the other; the top of the barrel body 1 is provided with a garbage inlet 11, the garbage classification unit 2 comprises a first-level conveyor belt 21, a compression mechanism 23 located below a first end of the first-level conveyor belt 21, a second-level conveyor belt 22 located below a second end of the first-level conveyor belt 21 and an image acquisition camera 6 located above the first-level conveyor belt 21, the first-level conveyor belt 21 is located below the garbage inlet 11, the conveying directions of the first-level conveyor belt 21 and the second-level conveyor belt 22 are arranged in a linear staggered mode, and the garbage classification collection unit 3 comprises a recoverable garbage basket 31 arranged below the compression mechanism 23, and a dry garbage basket 32 and a wet garbage basket 33 respectively arranged below the first end and the second end of the second-level conveyor belt 22.
More specifically, in the present embodiment:
the garbage classification unit 2 further comprises a controller 4, and the controller 4 is electrically connected with the first-level conveyor belt 21, the second-level conveyor belt 22, the compression mechanism 23 and the image acquisition camera 6 respectively. The image acquisition camera 6 can be selected from various commercially available cameras, and preferably adopts a color camera. The controller may be a commercially available ARM or X86 architecture controller or other suitable controller known in the art. The image acquisition camera 6 is used for acquiring image information of garbage falling on the primary conveyor belt 21, and the controller 4 is used for receiving the image information of the garbage, identifying the image information of the garbage based on deep learning, judging the garbage type, and further controlling the primary conveyor belt 21, the secondary conveyor belt 22 and/or the compression mechanism 23 to work so that the garbage enters a corresponding garbage basket.
An openable lid may also be provided at the waste inlet 11. In actual use, can also set up human sensor to be connected with controller 4, can realize that the automation of case lid is opened when sensing the user and being close to rubbish entry 11, the person of facilitating the use uses this garbage bin more.
The straight lines of the conveying directions of the primary conveyor belt 21 and the secondary conveyor belt 22 are preferably perpendicular to each other in the horizontal plane. As shown in (a) to (d) of fig. 4. More preferably, the primary conveyor belt 21 and the secondary conveyor belt 22 both have a belt 201 and a driving mechanism for driving the belt 201 to convey forward and backward, and the driving mechanism is composed of a driven wheel 203, a driving wheel 202 and a driving motor 204 for driving the driving wheel 202 to rotate forward and backward.
The compression mechanism 23 is preferably a can compression mechanism. The compression mechanism of the pop can select related products in the prior art. Or the zip-top can compressing mechanism is provided with a barrel with an upper opening and a lower opening, the bottom of the barrel is provided with a bottom baffle capable of being turned downwards, the inner side wall of the barrel is provided with a turnover compressing plate, the turnover compressing plate is driven by a cylinder arranged on the barrel in a penetrating manner to be matched with the bottom baffle, garbage is compressed after falling into the barrel, and after the compression is finished, the bottom baffle is opened to enable the garbage to fall into the recyclable garbage basket.
The classification method of the intelligent classification garbage can based on the deep learning is shown in fig. 5, and comprises the following steps:
s1: the waste falls onto the primary conveyor 21 via the waste inlet 11;
s2: the image acquisition camera 6 samples, gathers the image information of the rubbish that falls on the one-level conveyer belt 21, and the image information of rubbish is received to controller 4 to compare with the storehouse in, judge the rubbish type:
if the garbage is recyclable garbage, the primary conveyor belt 21 drives the garbage to be conveyed towards the first end of the primary conveyor belt, the garbage is conveyed into the compressing mechanism 23, the compressing mechanism 23 compresses the garbage, the garbage falls into the recyclable garbage basket 31 after the garbage is processed,
if the garbage is dry garbage, the primary conveyor belt 21 drives the garbage to be conveyed towards the second end direction, the garbage is conveyed to the secondary conveyor belt 22, the secondary conveyor belt 22 drives the garbage to be conveyed towards the first end direction, the garbage is conveyed into the dry garbage basket 32,
if the garbage is wet garbage, the primary conveyor belt 21 drives the garbage to be conveyed towards the second end direction of the primary conveyor belt, the garbage is conveyed onto the secondary conveyor belt 22, the secondary conveyor belt 22 drives the garbage to be conveyed towards the second end direction of the secondary conveyor belt, and the garbage is conveyed into a wet garbage basket 33;
and step S2 is carried out based on machine vision and by using a deep learning mode of a convolutional neural network.
Putting harmful garbage: when harmful garbage is thrown in, the garbage should be taken and put lightly; putting the waste medicines together with packages; harmful garbage such as pressure tanks is thrown after the content in the pressure tanks is emptied; when harmful garbage is generated in public places, the harmful garbage needs to be carried to a harmful garbage throwing point to be properly thrown; and for harmful garbage which is easy to damage, the garbage needs to be completely wrapped and then thrown.
In this embodiment, based on a deep learning algorithm:
(2) Current situation of deep learning
Deep learning is one of the new research directions in the field of machine learning. However, in most cases, machine learning can almost replace the concept of artificial intelligence. In short, deep learning is an algorithm using machine learning, so that a computer can learn intrinsic rules and characteristics in a database, so as to identify and predict other samples. Deep learning is a complex machine learning algorithm that achieves excellent results in many respects, and outperforms other related techniques in speech and image recognition. Typical deep learning modes include a convolutional neural network model, a deep trusted network model, a stack coordination network model and the like.
(2) The embodiment adopts the convolutional neural network and the intelligent classification garbage can based on machine vision
The convolutional neural network is a layered neural network consisting of alternating convolutional layers and sub-sampling layers and is used for simulating cells of visual nerves, and different models of the convolutional neural network are characterized by implementation modes of the convolutional layers and the sub-sampling layers and training modes of the convolutional layers and the sub-sampling layers.
Parameters of the convolutional layer include: the number of input images, the number of characteristic images, the size of the images and the size (Mx, my) of each layer of images are the same; the sizes (Kx, ky) of the convolution kernels, each of the sizes (Kx, ky) of the convolution kernels being an effective area to be applied to the input image; the skip factor (Sx, sy) defines how many pixels the convolution kernel skips in the x, y direction. The size of the output image obtained after feature extraction of the convolutional layer is obtained by the following formula:
the sampling layer construction method comprises the following steps: the biggest difference realized by the convolutional neural network in the system is that a maximum pool sampling layer is adopted to replace a sub-sampling layer. In the convolutional neural network implementation, these layers are replaced by pool sampling and averaging operations, and adjacent pixels are skipped during convolution for sampling purposes. The output of the maximum pool sampling layer is obtained by taking the maximum value of the matrices (Kx, ky) whose sizes do not overlap. The maximum pool sampling provides local displacement invariance, down-sampling each direction of the input image by a (Kx, ky) factor. In the experiment, the maximum pool sampling is carried out on the characteristic image area through a 2 multiplied by 2 filtering window, the maximum value of the window is extracted as the sampling characteristic, and the characteristic image is subjected to down-sampling. Construction of the classification layer: the convolution kernel size, the maximum pool sampling matrix and the jump factor of the convolution filter are selected, the output image of the last convolution layer is down-sampled to one pixel, and a full connection layer combines the output of the last convolution layer into a one-dimensional characteristic matrix. In the classification task, the last layer is usually a fully connected layer that connects each pixel image to every possible classification of the output layer. The final layer uses softmax regression as the activation function, and the output of each neuron represents the likelihood of the classification result.
(3) Image acquisition and processing
The items needing to be classified are photographed and collected, the same object is photographed at different angles in the same placing mode, and the same object is photographed at different placing modes and at multiple angles. The implementation of a convolutional neural network requires training images of the same size. Namely, the shot photo is cut into sizes, and the cut size is 4032 × 3024 pixels.
The shot photos are divided into three categories of dry garbage, wet garbage and recyclable garbage, and the photos are moved into three folders. Besides the three garbage photos to be identified, the photos of the conveyor belt are also acquired. Since the conveyor belt is used as the whole background in the recognition process, the photograph of the conveyor belt is collected so as to allow the computer to remove the background and then perform comparison in the recognition process.
Because the acquired training sample has obvious light change, the light change also influences the recognition accuracy in the actual test, so that the recognition is more accurate. The brightness of the image is adjusted to obtain sample pictures under different light rays.
(4) Model training
Model training this process is called transfer learning, which is commonly used in deep learning applications. A pre-trained network is used as a starting point for learning a new task.
Tuning the network through migratory learning is typically faster and easier than training a network of randomly initialized weights starting from zero. The learned functionality can be quickly transferred to a new task using a small number of training images. A pre-trained network is trained to classify new images, and to change the match data, the final layer is replaced with a new layer adapted to the new data set. The output size is compiled as the number of classes in the new data. The learning rate is compiled to learn faster in the new layer than in the transport layer, deleting the original layer and joining the new layer. The output layer is replaced, scrolled to the end of the "layer" palette, and the new sort output layer is dragged onto the canvas. The original output layer is deleted and the new layers are joined. After being ready for training, "analyze" is clicked and the error reported by the deep learning network analyzer is guaranteed to be zero. As shown in fig. 6.
Returning to the deep network designer, and clicking "export," the deep network designer exports the network into a new variable named lgraph _1, which contains the edited network layers. The layer variables can now be provided to the trainNetwork function. Uncompressing the new images and loading the images as an image data store, the imageDatastore may automatically tag the images according to the name of the folder and store the data as the object of the imageDatastore. In the training process of the convolutional neural network, a large amount of image data including data which cannot be stored in a memory can be stored, and images can be effectively read in batches.
Before training, the compressed file is decompressed, a folder is added to the directory, and a training sample is arranged in the folder. To organize data more easily, a data structure is constructed using the imagedatastore function to manage the data. The data is automatically divided into training data and validation data, wherein the training data accounts for 70% and the validation data accounts for 30%. The image size needs to be adjusted to match the input size of the pre-trained network.
Specifying a training option: a small batch size is specified, i.e. how many images to use per iteration. A small number of epochs are specified, with epochs being a complete training period over the entire training data set. For transfer learning, training is not required for as many epochs, and the data for each epoch is reshuffled. The initial learning rate is set to a small value to slow down the learning speed in the transport layer. Specifying authentication data and a small authentication frequency. And controlling to open a training scenario and monitoring the progress during training.
To train the network, the trainNetwork function is provided with layers, training images and options derived from the application lgraph _ 1. By default, the rainnetwork uses an available GPU (requiring parallel computing tools), otherwise it uses a CPU. The training process is shown in fig. 7, 8, and 9.
The training time can be obtained by the training result for 32 minutes, and the accuracy of image recognition after training reaches more than 90%.
(5) Garbage sorting test
Simulating garbage sorting through a computer. And classifying the verification images by using a fine adjustment network, and calculating the classification precision. Four pictures are extracted from the folder at one time in an arrangement mode of 2 x 2, and four sample verification images with prediction labels are displayed on a computer window after simulated sorting. The garbage can is used for testing garbage sorting, the trained model is selected, garbage is placed in the garbage can, a photo is obtained through a camera on the garbage can, and the size of the photo is modified according to the trained model. And inquiring the category, and displaying the photo and the garbage category on a computer window.
The embodiments described above are intended to facilitate the understanding and use of the invention by those skilled in the art. It will be readily apparent to those skilled in the art that various modifications to these embodiments may be made, and the generic principles described herein may be applied to other embodiments without the use of the inventive faculty. Therefore, the present invention is not limited to the above embodiments, and those skilled in the art should understand that all the improvements and modifications made without departing from the scope of the present invention according to the disclosure of the present invention should be within the protection scope of the present invention.
Claims (3)
1. A classified garbage can comprises a can body (1), a garbage classifying unit (2) and a garbage classifying and collecting unit (3), wherein the garbage classifying unit and the garbage classifying and collecting unit are arranged in the can body (1) from top to bottom;
the top of the barrel body (1) is provided with a garbage inlet (11),
the garbage classification unit (2) comprises a primary conveyor belt (21), a compression mechanism (23) positioned below the first end of the primary conveyor belt (21) and a secondary conveyor belt (22) positioned below the second end of the primary conveyor belt (21), the primary conveyor belt (21) is positioned below the garbage inlet (11), the conveying directions of the primary conveyor belt (21) and the secondary conveyor belt (22) are arranged in a linear staggered manner,
the garbage classification and collection unit (3) comprises a recyclable garbage basket (31) arranged below the compression mechanism (23) and a dry garbage basket (32) and a wet garbage basket (33) which are respectively arranged below the first end and the second end of the secondary conveyor belt (22).
2. A sorting container according to claim 1, characterized in that the garbage sorting unit further comprises a control unit, the control unit comprises a controller (4) and a garbage sorting button assembly (5), the controller (4) is electrically connected to the primary conveyor (21), the secondary conveyor (22), the compressing mechanism (23) and the garbage sorting button assembly (5), respectively, and the garbage sorting button assembly (5) comprises a wet garbage button, a dry garbage button and a recyclable garbage button for controlling garbage entering the garbage inlet (11) to enter the corresponding garbage basket of the garbage sorting and collecting unit (3).
3. A sorting bin according to claim 1, characterised in that it includes any one or more of the following conditions:
a. the straight lines of the conveying directions of the primary conveyor belt (21) and the secondary conveyor belt (22) are vertical to each other in the horizontal plane;
b. the primary conveyor belt (21) and the secondary conveyor belt (22) are respectively provided with a belt (201) and a driving mechanism for driving the belt (201) to convey forward and backward, and the driving mechanism consists of a driven wheel (203), a driving wheel (202) and a driving motor (204) for driving the driving wheel (202) to rotate forward and backward;
c. the compression mechanism (23) is a zip-top can compression mechanism;
d. an openable box cover is arranged at the garbage inlet (11).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202121481089.XU CN217576626U (en) | 2021-06-30 | 2021-06-30 | Classification garbage can |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202121481089.XU CN217576626U (en) | 2021-06-30 | 2021-06-30 | Classification garbage can |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN217576626U true CN217576626U (en) | 2022-10-14 |
Family
ID=83526372
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202121481089.XU Active CN217576626U (en) | 2021-06-30 | 2021-06-30 | Classification garbage can |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN217576626U (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113371363A (en) * | 2021-06-30 | 2021-09-10 | 上海建桥学院有限责任公司 | Classified garbage can, intelligent classified garbage can based on deep learning and classification method |
-
2021
- 2021-06-30 CN CN202121481089.XU patent/CN217576626U/en active Active
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113371363A (en) * | 2021-06-30 | 2021-09-10 | 上海建桥学院有限责任公司 | Classified garbage can, intelligent classified garbage can based on deep learning and classification method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109606991B (en) | Intelligent garbage can and garbage classification method based on deep learning | |
| CN110861852A (en) | Automatic classification garbage can based on visual identification and classification method | |
| CN110053903A (en) | Categorized consumer waste retrieval management system and method | |
| CN110697273A (en) | Intelligent household garbage identification and automatic classification system and method based on iterative learning control | |
| CN113371363A (en) | Classified garbage can, intelligent classified garbage can based on deep learning and classification method | |
| CN106904386A (en) | A kind of automatic classification dustbin based on mode identification technology | |
| CN110577037A (en) | Method for classifying, checking and recycling household garbage | |
| CN111559588A (en) | Intelligent garbage can for classified garbage throwing and classified garbage throwing method | |
| CN112827846B (en) | Automatic garbage classification device and method | |
| CN217576626U (en) | Classification garbage can | |
| CN111056191A (en) | Intelligent garbage collection system capable of classifying and delivering multiple garbage | |
| CN116682098A (en) | Automatic urban household garbage identification and classification system and method | |
| CN110929693A (en) | Intelligent garbage classification method and system | |
| CN110654740A (en) | Method, device and storage medium for automatically classifying garbage | |
| CN110342133A (en) | A kind of dustbin and packing method of automatic packaging waste | |
| Sirawattananon et al. | Designing of IoT-based smart waste sorting system with image-based deep learning applications | |
| CN114955289A (en) | Intelligent garbage classification recycling and management method and intelligent classification garbage can | |
| KR20210074929A (en) | deep learning based garbage distribution system | |
| CN113859803A (en) | Intelligent identification trash can and intelligent identification method thereof | |
| CN214297580U (en) | Intelligent garbage recognition and classification device based on internet | |
| KR102412255B1 (en) | Recycling waste separation system | |
| CN116620744A (en) | Intelligent classification system based on deep learning and implementation method thereof | |
| CN116510901A (en) | Intelligent household garbage sorting system | |
| CN215708916U (en) | Intelligent recognition garbage can | |
| CN212607180U (en) | Triangular intelligent garbage can |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GR01 | Patent grant | ||
| GR01 | Patent grant |