TWI707812B - Smart resource recycling bin - Google Patents

Abstract

A smart resource recycling bin, including: a trash can, a plurality of cameras, and a neural network; wherein the cameras can photograph the items thrown into the trash can, and the neural network can receive the images captured by the cameras , Analyze and compare the image with the image stored in a database to determine the type of the item, and then control the electronic switch of an inner barrel according to this category to open the outer cover of the inner barrel, and Make the item fall into the corresponding inner bucket, so that the items thrown into the trash can will be automatically sorted and dropped into the corresponding inner bucket, achieving the efficiency and purpose of automatically judging the sorting resource recovery .

Description

Smart resource recycling bin

The present invention is a smart resource recycling bin, especially a smart resource recycling bin that can automatically determine and classify garbage.

According to the economic growth, the standard of living and quality have improved, and the amount of garbage produced in people’s lives has also increased. However, there are still many recyclable resources in all kinds of garbage. Such as iron, aluminum cans, plastic bottles and paper, etc. Especially in recent years, the topic of environmental protection has been given considerable attention. The importance of resource recycling has always been a topic that governments around the world have been vigorously promoting. Therefore, it has been promoted in public places, such as outdoors, schools, playgrounds, etc., or in general households. Place recycling bins to provide different types of recycling materials.

Take the common resource recycling bin in public places as an example. Please refer to Figure 1. It has a metal barrel 91 made of metal material, and a metal barrel cover 92 arranged above the metal barrel 91. , And a plurality of inner barrels 93 that can be accommodated inside the metal barrel 91. In this way, it is provided that the user can put the recyclables into the inner barrel 93 located inside the metal barrel 91 through the metal barrel cover 92 according to the classification, so as to achieve the purpose of resource recovery.

However, the conventional resource recycling bin is manually judged by the type of recyclable, and then put into the corresponding inner barrel 93 after the judgment is made. If someone throws the inner barrel 93 in a litter or wrong, it will cause subsequent The difficulty of sorting out is quite inconvenient.

It can be seen that there are still many shortcomings in the above-mentioned conventional items, and they are not a good designer, and urgently need to be improved.

In view of this, the inventor of this case has been engaged in the manufacturing, development and design of related products for many years. Aiming at the above-mentioned goals, after detailed design and careful evaluation, he finally obtained a practical invention.

The purpose of the present invention is to provide a smart resource recycling bin, which automatically judges and recycles plastic, iron and aluminum cans, glass, aluminum foil bags and other garbage.

According to the above-mentioned purpose, the smart resource recycling bin of the present invention mainly includes: a trash bin, a plurality of cameras and a neural network; wherein the trash bin is provided with a plurality of inner barrels, and each of the inner barrels is provided for separately Different types of resource recycling, each of the openings of the inner bucket is provided with an outer cover, and each outer cover is provided with an electronic switch, so that each of the outer covers can be opened by controlling each of the electronic switches; the cameras are installed in the trash can Around the top of the edge, it shoots images of various recyclable items and other garbage; the neural network is set on the trash can, and is electrically connected to the electronic switches and cameras. The neural network has a data The library, which receives the images captured by the cameras, learns the images of various recyclable items, and stores the learned images in the database; thereby, the cameras can shoot the Throw the object into the trash can, make the neural network receive the image captured by the cameras, compare the image with the image stored in the database, analyze and compare the object to determine the type of the object, and then follow this The category control corresponds to the electronic switch of the inner bucket, so that the outer cover of the inner bucket is opened, and the item falls into the corresponding inner bucket. In this way, the items thrown into the trash can are automatically classified and Fall into the corresponding inner bucket to achieve the purpose of automatically judging and recycling resources.

In order to facilitate your reviewer to have a further understanding and understanding of the purpose, shape, features and effects of the structure of the present invention, the examples and diagrams are described in detail as follows:

The present invention relates to a "smart resource recycling bin". Please refer to Figures 2, 3, and 4. The smart resource recycling bin of the present invention mainly includes: a trash can 10, a plurality of cameras 20 and a nerve Network 30.

Wherein, the trash can 10 is provided with a plurality of inner barrels 11, and each of the inner barrels 11 is provided for different types of resource recycling. The opening of each inner barrel 11 is provided with an outer cover 12, and each outer cover 12 has an electronic The switch 13 allows each of the electronic switches 13 to be controlled to open each of the outer covers 12.

The cameras 20 are arranged around the top of the inner edge of the trash can 10, and shoot images 21 of various recyclable items and other trash.

The neural network 30 is set on the trash can 10 and is electrically connected to the electronic switches 13 and the cameras 20. The neural network 30 is provided with a database 31 that receives the images 21 captured by the cameras 20 , And learn the images 21 of various recyclable items, and store the learned images 21 in the database 31.

With the composition of the above-mentioned components, during resource recovery, the cameras 20 can photograph the objects thrown into the trash can 10, so that the neural network 30 receives the images 21 captured by the cameras 20, and combines the images 21 with The image 21 stored in the database 31 is analyzed and compared 32 to determine the category 22 of the item, and then the electronic switch 13 corresponding to the inner barrel 11 is controlled according to the category 22, so that the outer cover 12 of the inner barrel 11 is opened , And make the item fall into the corresponding inner barrel 11.

In this way, the items thrown into the trash can 10 can be automatically sorted and dropped into the corresponding inner bucket 11, so that people who don’t have time to sort can throw the trash directly without worrying about what will happen when littering. This problem can reduce the working time of recycling personnel, not only can effectively recycle and sort and reduce the amount of garbage, but also can reduce the damage to the ecological environment, and achieve the efficiency of automatic judgment and classification of resource recovery, and through convenient and fast classification The effect makes it easy for the public to classify and achieve the purpose of caring for the earth.

Please refer to Figures 2, 3 and 4. The neural network 30 controls the electronic switch 13 corresponding to the inner barrel 11. Signal 1 is to open the outer lid 12, and signal 0 is to close the outer lid 12 (as shown in Figure 3). Shown).

Please refer to Figures 2, 3, and 4 again. The neural network 30 utilizes an electric field programmable logic gate array (FPGA) for high-speed operations to achieve rapid analysis.

Please refer to Figures 2, 3, and 4 again. The category 22 of this article is divided into 4 categories, 1 is PET bottles (plastic), 2 is iron and aluminum cans, and 3 is aluminum foil bags. The fourth is garbage.

Please refer to Figures 2, 3, and 4. This neural network 30 is a mathematical model that imitates the structure and function of a biological neural network and is used to estimate or approximate functions. It is composed of a large number of artificial neurons. Connected for calculation, the artificial neural network can change the internal structure on the basis of external information, has a learning function, is a non-linear statistical framework, and is optimized by a learning method based on mathematical statistics. , Obtain a large number of local structural spaces that can be expressed by functions, and make artificial perception decisions through the application of mathematical statistics.

Please refer to Figures 2, 3, and 4 again. The image 21 stored in the database 31 is designed in an object-oriented manner, including the attributes, methods, and categories of the objects.

Please refer to Figures 2, 3, and 4. The neural network 30 is a supervised learning method, that is, learning through labeled training data (with a supervisor to guide the learning), and an artificial neural network contains multiple layers of nodes ; The connection between the input layer, the intermediate hidden layer and the adjacent layer nodes of the output layer all have "weights". The purpose of learning is to assign these edges to the correct weights. Through the input vector, these weights can determine the output vector, all edges Edge weights are randomly assigned pairs, and for all the inputs in the training data set, they are activated to observe their outputs. These outputs will be compared with the correct output, and the error will be passed back to the previous layer. The error will be marked, and the weight will be adjusted accordingly. The process is repeated until the output error is lower than the established standard.

Please refer to Figures 2, 3, and 4 again. The electronic switch 13 can be a stepping motor module, and the neural network 30 controls the stepping motor module to actuate and open and close the inner barrel 11 Cover 12.

Please refer to Figures 2, 3, and 4 again. There are four inner barrels 11 for collecting and sorting PET bottles, iron and aluminum cans, aluminum foil bags, garbage and other resource recycling items.

Please refer to Figs. 2, 3, and 4 again, the lens of the cameras 20 can be a wide-angle lens, so that a wide-angle image 21 can be taken.

Please refer to Figures 2, 3, and 4 again. The cameras 20 are provided with three cameras to respectively photograph the four sides (front, back, and sides) of the objects thrown into the trash can 10.

Please refer to Figures 2, 3, and 4 again. The neural network 30 receives the front, back, and side images 21 of the object captured by the three cameras 20, and stores the images 21 in the database 31. The image 21 is analyzed and compared 32 to determine the category 22 of the item. The method for the analysis and comparison 32 is to use the three cameras 20 to capture the image 21, that is, the three sets of images 21 each determine the category 22 of the item It belongs to PET bottles, iron and aluminum cans, aluminum foil bags, or garbage, and judges the category 22 of the item with the more category 22.

That is, the image 21 captured by two cameras 20 judged (determined above 50%) belongs to the PET bottle category, while the image 21 captured by the third camera 20 judged (determined above 50%) belongs to For iron and aluminum cans, by voting, PET bottles are selected as the more judged category 22, and the item category 22 is judged to be PET bottles.

In summary, the smart resource recycling bin of the present invention does have an unprecedented innovative structure. It has not been seen in any publications, and there is no similar product on the market. Therefore, its novelty should be considered. In addition, the unique features and functions of the present invention are far from comparable with conventional ones, so it is indeed more progressive than conventional ones, and it complies with the requirements of the Chinese Patent Law regarding the requirements for invention patent applications.

The above are only the best specific embodiments of the present invention, but the structural features of the present invention are not limited thereto. Any change or modification that can be easily thought of by anyone familiar with the art in the field of the present invention can be covered In the following patent scope of this case.

10: trash can 11: inner barrel 12: Outer cover 13: Electronic switch 20: Camera 21: Image 22: Category 30: Neural Network 31: Database 32: Analysis and comparison

Figure 1 is a three-dimensional schematic diagram of a conventional recycling bin.

Figure 2 is a block diagram of the smart resource recycling bin of the present invention.

Figure 3 is a flow chart of the operation of the smart resource recycling bin of the present invention.

Figure 4 is a schematic diagram of the appearance of the trash bin of the smart resource recycling bin of the present invention.

10: trash can

20: Camera

30: Neural Network

31: Database

Claims (10)

A smart resource recycling bin, including: A trash can. The trash can is provided with multiple layers of inner buckets. Each of the inner buckets is used for different types of resource recycling. The bottom inner bucket is for other garbage. The opening of each inner bucket is provided with one The outer cover, each of the outer covers has an electronic switch, so that each of the outer covers can be opened by controlling each of the electronic switches; A plurality of cameras, which are set around the top of the inner edge of the trash can, and shoot images of various recyclable items and other garbage; and A neural network is set on the trash can and is electrically connected to each of the electronic switches and the cameras. The neural network has a database that receives the images captured by the cameras and learns Images of various recyclable items, and store the images after learning in the database; Among them, during resource recovery, the cameras can take pictures of the items thrown into the trash can, so that the neural network can receive the images captured by the cameras, and analyze the images with the images stored in the database The category of the item is determined by comparison, and then the electronic switch of the corresponding inner barrel is controlled according to the category, so that the outer cover of the inner barrel is opened, and the article falls into the corresponding inner barrel. For example, in the smart resource recycling bucket described in item 1 of the scope of patent application, the neural network controls the electronic switch corresponding to the inner bucket. Signal 1 is to open the outer lid, and signal 0 is to close the outer lid. For example, in the smart resource recycling bin described in item 2 of the scope of patent application, the neural network utilizes an electric field programmable logic gate array (FPGA) for high-speed calculation and rapid analysis. For example, the smart resource recycling bin described in item 1 of the scope of patent application, in which the categories of the items are divided into 4 categories, 1 is PET bottles (plastic), 2 is iron and aluminum cans, and 3 is aluminum foil bag Category, 4 is garbage. For example, the smart resource recycling bin described in item 1 of the scope of patent application, in which the images stored in the database are designed in an object-oriented manner, including the attributes, methods, and categories of the objects. For example, the intelligent resource recycling bucket described in the first item of the scope of patent application, wherein the electronic opening relationship is a stepping motor module, and the neural network controls the stepping motor module to operate, and opening and closing the inner bucket cover. For example, the smart resource recycling bin described in item 4 of the scope of patent application has four inner bins for collecting and sorting PET bottles, iron and aluminum cans, aluminum foil bags, and garbage. For example, in the smart resource recycling bin described in item 1 of the scope of patent application, the lens of the cameras is a wide-angle lens. For example, in the smart resource recycling bin described in item 8 of the scope of patent application, the cameras are equipped with three cameras to take pictures of the front, back and sides of the items thrown into the trash bin. For example, the smart resource recycling bucket described in item 9 of the scope of patent application, wherein the neural network receives the front, back, and side images of the object captured by the three cameras, and combines these images with the images stored in the database, Analyze and compare to determine the category of the item. The method for analysis and comparison is to use the three cameras to capture images, that is, three sets of images to determine whether the category of the item belongs to PET bottles, iron and aluminum cans. , Aluminum foil bags, or garbage, and judge the category of the item by judging more categories.

Images