Large garbage treatment method and system

Abstract

The embodiment of the invention relates to the technical field of garbage treatment, and discloses a large garbage treatment method and system, and the method comprises the steps: obtaining a large garbage image corresponding to large stacked garbage at a large garbage collection part in real time through a first intelligent monitoring unit; according to the large garbage images, large garbage data at the large garbage recycling position are analyzed and counted in real time; according to the large garbage data and the vehicle information, a loading model is used for prediction, and the number of to-be-loaded vehicles and a large garbage loading scheme correspondingly loaded by each to-be-loaded vehicle are determined, so that a user calls the to-be-loaded vehicles to load the corresponding large garbage; and the second intelligent monitoring unit is used for acquiring a large garbage sorting video in real time, identifying and detecting the large garbage sorting video, determining whether a sorting error exists in the large garbage sorting process or not, and sending a sorting error prompt to a user in real time. According to the embodiment of the invention, the effects of optimizing the garbage treatment process and improving the efficiency are achieved.

Description

Technical Field

[0001] The embodiments of the present invention relate to the technical field of garbage disposal, and in particular to a method and system for disposing large garbage. Background Art

[0002] At present, with the renewal and development of cities and the improvement of residents' living standards, the amount of large-sized garbage is on the rise.

[0003] However, large pieces of garbage are large in size, heavy in weight, and occupy a lot of space. They are inconvenient to collect and transport, and the transportation cost is high. In addition, due to the characteristics of volume and weight, large pieces of garbage have high transportation costs and are inconvenient to disassemble on site. Generally, they are transported from garbage collection points to sorting and processing centers for disassembly and classification.

[0004] The inventor of this application found that: on the one hand, there is a lack of reasonable planning for the stacking, collection, transportation, and disposal of large wastes, and in the process of transporting large wastes, due to the lack of experience of different workers in the classification and transportation of large wastes, there are unreasonable loading and ineffective use of vehicle resources during the transportation of large wastes, resulting in high transportation costs and low efficiency. On the other hand, there are classification errors when sorting large wastes, resulting in a low recycling rate of large waste resources. Summary of the invention

[0005] In view of the above problems, an embodiment of the present invention provides a large-scale garbage disposal method, which is used to solve the problems of unreasonable loading, unreasonable vehicle resource scheduling, low resource recovery rate and low garbage disposal efficiency existing in the prior art.

[0006] According to one aspect of an embodiment of the present invention, a method for processing large-scale garbage is provided. The method is based on a large-scale garbage processing system, and the large-scale garbage processing system includes a video monitoring device; the video monitoring device includes at least one first intelligent monitoring unit arranged at a large-scale garbage collection location, and at least one second intelligent monitoring unit arranged at a large-scale garbage sorting and processing point. The method includes:

[0007] Acquire, in real time, through the first intelligent monitoring unit, a large-scale garbage image corresponding to the large-scale accumulated garbage at the large-scale garbage collection site;

[0008] According to the large-scale garbage image, the large-scale garbage data of the large-scale garbage collection site is analyzed and counted in real time; the large-scale garbage data includes the number of large-scale garbage, the type of large-scale garbage and the basic attribute information of large-scale garbage; the basic attribute information of large-scale garbage includes the size information and weight information of large-scale garbage;

[0009] According to the large-scale garbage data and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded and the large-scale garbage loading plan corresponding to each vehicle to be loaded, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage; the loading model is obtained by training a deep reinforcement learning algorithm using an objective function and constraints based on the large-scale garbage data samples and vehicle information; the objective function is used to maximize the space utilization of the vehicle; the constraints include the garbage type constraints, size constraints, weight constraints of the large-scale garbage and the space constraints of the vehicle; the large-scale garbage loading plan includes the relative positions of each large-scale garbage to be loaded in the same vehicle to be loaded;

[0010] During the sorting process at the large-scale garbage sorting and processing point, the large-scale garbage sorting video is obtained in real time through the second intelligent monitoring unit, and the large-scale garbage sorting video is identified and detected to determine whether there is a sorting error in the large-scale garbage sorting process. When a sorting error occurs, a sorting error prompt is sent to the user in real time.

[0011] In an optional manner, before analyzing and collecting the large-scale garbage data of the large-scale garbage collection site in real time according to the large-scale garbage image, the method further comprises:

[0012] Obtaining a large garbage image sample; the large garbage image sample includes a large garbage sample image and a corresponding sample label; the sample label includes a large garbage type label, a large garbage size label, and a weight label corresponding to the large garbage sample image;

[0013] Input the large garbage image sample into a preset YOLOV5 model for training to obtain a trained large garbage type recognition model; the large garbage type recognition model is used to identify large garbage, large garbage type and large garbage size in the large garbage image;

[0014] The recognition result of the large-scale garbage type recognition model on the large-scale garbage image sample is input into a preset deep learning model for training to obtain a trained large-scale garbage weight prediction model; the large-scale garbage weight prediction model is used to predict the weight information of large-scale garbage.

[0015] In an optional manner, the real-time analysis and statistics of the bulky waste data at the bulky waste collection site based on the bulky waste image includes:

[0016] By inputting the large-scale garbage image into a large-scale garbage type recognition model in real time, the large-scale garbage type and large-scale garbage size information of each large-scale garbage in the large-scale garbage image are identified;

[0017] Inputting the large-piece garbage type and large-piece garbage size information of each large-piece garbage into the large-piece garbage weight prediction model for prediction to obtain the weight information of each large-piece garbage;

[0018] The number of large-sized garbage in each of the large-sized garbage images, the corresponding large-sized garbage type, the large-sized garbage size information and the weight information are counted in real time, so as to obtain the large-sized garbage data of the large-sized garbage recycling site.

[0019] In an optional manner, the types of large garbage include at least one or more of complete and undeformed large garbage, damaged metal large garbage, deformed metal large garbage, damaged fabric large garbage, deformed fabric large garbage, damaged wooden large garbage, deformed wooden large garbage, damaged wooden large garbage, and deformed wooden large garbage.

[0020] In an optional manner, the prediction is performed using a loading model based on the large-scale garbage data and vehicle information to determine the number of vehicles to be loaded and the large-scale garbage information corresponding to each vehicle to be loaded, so that the user calls the vehicle to be loaded to load the corresponding large-scale garbage, including:

[0021] According to the complete and undeformed large-sized garbage and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded with the complete and undeformed large-sized garbage and the first large-sized garbage loading plan;

[0022] According to the damaged large-sized metal garbage, deformed large-sized metal garbage, damaged large-sized fabric garbage, deformed large-sized fabric garbage, damaged large-sized wooden garbage, deformed large-sized wooden garbage, damaged large-sized wooden garbage, deformed large-sized wooden garbage and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded and the loading plan for the second largest garbage;

[0023] According to the first large-piece garbage loading plan, the intact and undeformed large-piece garbage is loaded on the same vehicle to be loaded;

[0024] According to the second large-piece garbage loading plan, the damaged large-piece metal garbage, deformed large-piece metal garbage, damaged large-piece fabric garbage, deformed large-piece fabric garbage, damaged large-piece wooden garbage, deformed large-piece wooden garbage, damaged large-piece wooden garbage, and deformed large-piece wooden garbage are respectively loaded onto the corresponding vehicles to be loaded.

[0025] In an optional manner, at least one illegal loading camera device is further provided above the loading space of the vehicle to be loaded; during the process of loading large garbage into the vehicle to be loaded, the method further comprises:

[0026] The illegal loading camera device acquires loading images during the loading process in real time;

[0027] Determine whether there is any non-compliant garbage on the vehicle to be loaded according to the loading image; wherein the non-compliant garbage at least includes: whether other garbage except large garbage is illegally loaded.

[0028] In an optional manner, during the sorting process at the large-scale waste sorting processing point, the large-scale waste sorting video is obtained in real time by the second intelligent monitoring unit, and the large-scale waste sorting video is identified and detected to determine whether there is a sorting error in the large-scale waste sorting process. When there is a sorting error, before sending a sorting error prompt to the user in real time, the method further includes:

[0029] Obtaining information on each large-scale waste sorting and processing point and information on second-hand recycling points within the target range; the information on the large-scale waste sorting and processing point includes location information and current working status information of the large-scale waste sorting and processing point;

[0030] Generating first path planning information for a vehicle loaded with intact and undeformed large garbage based on the information of the second-hand recycling point;

[0031] According to the information of each large-scale garbage sorting and processing point, the large-scale garbage sorting and processing points corresponding to the remaining loading vehicles are determined, and the second path planning information of the path is generated respectively.

[0032] In an optional manner, during the sorting process at the large-scale waste sorting processing point, the large-scale waste sorting video is obtained in real time by the second intelligent monitoring unit, and the large-scale waste sorting video is identified and detected to determine whether there is a sorting error in the large-scale waste sorting process. When a sorting error occurs, a sorting error prompt is sent to the user in real time, including:

[0033] The large-piece garbage sorting video is input into the garbage sorting detection model to determine whether there is large-piece garbage that is sorted incorrectly in each sorting area; wherein the garbage sorting detection model is pre-trained based on the sorting video samples input into the neural network model, and the sorting video samples include the sorting sample videos and the sorting garbage type labels of each sorting area in the large-piece garbage sorting processing point in the sorting sample videos; the sorting garbage labels include metal parts, wooden parts, and disassembled mixed garbage.

[0034] According to another aspect of an embodiment of the present invention, a large-scale garbage processing system is provided, the system comprising a video monitoring device; the video monitoring device comprises at least one first intelligent monitoring unit arranged at a large-scale garbage collection location, at least one second intelligent monitoring unit arranged at a large-scale garbage sorting and processing point, an analysis and statistics unit, a loading prediction unit and a sorting detection unit; wherein:

[0035] The first intelligent monitoring unit is used to obtain in real time the large-scale garbage images corresponding to the large-scale garbage accumulation at the large-scale garbage collection site;

[0036] An analysis and statistics unit is used to analyze and count the large-scale garbage data of the large-scale garbage collection site in real time according to the large-scale garbage image; the large-scale garbage data includes the number of large-scale garbage, the type of large-scale garbage and the basic attribute information of large-scale garbage; the basic attribute information of large-scale garbage includes the size information and weight information of large-scale garbage;

[0037] A loading prediction unit is used to predict using a loading model based on the large-scale garbage data and vehicle information, determine the number of vehicles to be loaded and the large-scale garbage loading plan corresponding to each vehicle to be loaded, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage; the loading model is obtained by training a deep reinforcement learning algorithm based on the large-scale garbage data samples and vehicle information using an objective function and constraints; the objective function is used to maximize the space utilization of the vehicle; the constraints include the garbage type constraint, size constraint, weight constraint of the large-scale garbage and the space constraint of the vehicle; the large-scale garbage loading plan includes the relative position of each large-scale garbage to be loaded in the same vehicle to be loaded;

[0038] The sorting detection unit is used to obtain the large-piece garbage sorting video in real time through the second intelligent monitoring unit during the sorting process at the large-piece garbage sorting processing point, and identify and detect the large-piece garbage sorting video to determine whether there is a sorting error in the large-piece garbage sorting process, and when there is a sorting error, send a sorting error prompt to the user in real time.

[0039] In an optional manner, the system further comprises:

[0040] An image sample acquisition unit is used to acquire a large-piece garbage image sample; the large-piece garbage image sample includes a large-piece garbage sample image and a corresponding sample label; the sample label includes a large-piece garbage type label, a large-piece garbage size label and a weight label corresponding to the large-piece garbage sample image;

[0041] A large-scale garbage type recognition model training unit is used to input the large-scale garbage image sample into a preset YOLOV5 model for training to obtain a trained large-scale garbage type recognition model; the large-scale garbage type recognition model is used to identify large-scale garbage, large-scale garbage type and large-scale garbage size in the large-scale garbage image;

[0042] The large-scale garbage weight prediction model training unit is used to input the recognition result of the large-scale garbage type recognition model on the large-scale garbage image sample into a preset deep learning model for training to obtain a trained large-scale garbage weight prediction model; the large-scale garbage weight prediction model is used to predict the weight information of large-scale garbage.

[0043] The embodiment of the present invention determines the large-scale garbage data such as the quantity, type and basic attribute information of the large-scale garbage by setting up multiple video monitoring devices at the large-scale garbage recycling point and each large-scale garbage sorting and processing point through real-time analysis and statistics of the large-scale garbage images, and determines the required number of vehicles and the loading plan of each vehicle to be loaded based on these large-scale garbage data and vehicle information, so that the large-scale garbage can be pre-classified before being loaded into the vehicle, and the number of vehicles and loading space can be reasonably used. Furthermore, sorting errors can be identified by a sorting detection model at each large-scale garbage sorting and processing point, which can improve resource utilization. Through the above scheme, the processing efficiency of large-scale garbage can be effectively improved and the transportation cost can be reduced.

[0044] The above description is only an overview of the technical solution of the embodiment of the present invention. In order to more clearly understand the technical means of the embodiment of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings. In the accompanying drawings:

[0046] Figure 1 A schematic diagram of a process of a large-sized garbage treatment method provided by an embodiment of the present invention is shown;

[0047] Figure 2 A schematic diagram showing an application environment of a method for processing large-sized garbage provided by an embodiment of the present invention is shown;

[0048] Figure 3 A schematic structural diagram of a large-scale garbage disposal system provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0049] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0050] Figure 1 The flowchart of the method for handling large-sized garbage provided by an embodiment of the present invention is shown, and the method is executed by a large-sized garbage handling system. The large-sized garbage handling system includes a video monitoring device; the video monitoring device includes at least one first intelligent monitoring unit arranged at a large-sized garbage collection location, at least one second intelligent monitoring unit arranged at a large-sized garbage sorting and processing point, an analysis and statistics unit, a loading prediction unit, a sorting detection unit, and a vehicle to be loaded. Figure 1 As shown, the method comprises the following steps:

[0051] Step 110: Obtain, in real time, through the first intelligent monitoring unit, a large-scale garbage image corresponding to the large-scale accumulated garbage at the large-scale garbage recycling site.

[0052] Among them, Figure 2 As shown, the large-scale garbage collection area is used by residents to place large-scale garbage. A plurality of first intelligent monitoring units are arranged at the large-scale garbage collection area, and the first intelligent monitoring unit is a camera device. In the embodiment of the present invention, the image of large-scale garbage at the large-scale garbage collection area is obtained in real time. The large-scale garbage includes large-scale metal garbage, large-scale fabric garbage, large-scale wooden garbage, and large-scale mixed garbage.

[0053] Step 120: Analyze and collect statistics of the large-scale garbage data at the large-scale garbage collection site in real time based on the large-scale garbage image.

[0054] The large-sized garbage data includes the number of large-sized garbage, the type of large-sized garbage and the basic attribute information of large-sized garbage; the basic attribute information of large-sized garbage includes the size information and weight information of large-sized garbage. The types of large-sized garbage include at least one or more of complete and undeformed large-sized garbage, defective metal large-sized garbage, deformed metal large-sized garbage, defective fabric large-sized garbage, deformed fabric large-sized garbage, defective wooden large-sized garbage, deformed wooden large-sized garbage, defective wooden large-sized garbage and deformed wooden large-sized garbage. In the embodiment of the present invention, since the large-sized garbage of complete and undeformed electrical appliances is generally functionally damaged but not necessarily in appearance, the large-sized garbage of complete and undeformed electrical appliances in the embodiment of the present invention does not include large-sized garbage of electrical appliances.

[0055] In the embodiment of the present invention, the algorithm can be used to detect the large-sized garbage in the process of transporting and placing in real time, and count the number and type of large-sized garbage. Among them, the embodiment of the present invention adopts a large-sized garbage type recognition model to identify the large-sized garbage, large-sized garbage type and large-sized garbage size in the large-sized garbage image, and predicts the weight information of the large-sized garbage through a large-sized garbage weight prediction model.

[0056] Before this, the model needs to be trained in advance to obtain a large-scale garbage type recognition model and a large-scale garbage weight prediction model. Specifically, the following steps are included:

[0057] Step 001: Obtain large garbage image samples. The large garbage image samples include large garbage sample images and corresponding sample labels; the sample labels include large garbage type labels, large garbage size labels and weight labels corresponding to the large garbage sample images. In the embodiment of the present invention, large garbage type labels include large garbage name labels, complete and undeformed large garbage type labels, defective metal large garbage type labels, deformed metal large garbage type labels, defective fabric large garbage type labels, deformed fabric large garbage type labels, defective wooden large garbage type labels, deformed wooden large garbage type labels, defective wooden large garbage type labels, and deformed wooden large garbage type labels.

[0058] Step 002: Input the large garbage image samples into a preset YOLOV5 model for training to obtain a trained large garbage type recognition model.

[0059] Among them, the large garbage type recognition model is used to identify the large garbage name, large garbage type and large garbage size in the large garbage image. Since the large garbage type already contains material information, there is no need to set a separate material label.

[0060] In an embodiment of the present invention, a large garbage image sample is input into a preset YOLOV5 model to obtain a prediction result, and a loss function is calculated based on the prediction result and the sample label. When the loss function is greater than a preset loss threshold or the number of iterative training is less than a preset number of iterations, the parameters of the YOLOV5 model are adjusted, and the large garbage image sample is continuously input into the preset YOLOV5 model for training, and the loss is calculated until the loss is less than or equal to the preset loss threshold, or the preset number of iterative training is reached, and a trained large garbage type recognition model is obtained. In an embodiment of the present invention, a classification loss function can be used to calculate the loss of YOLOV5, and specifically, a cross entropy loss function can be used for calculation.

[0061] Among them, the cross entropy loss function is:

[0062]

[0063] in: is the value of the loss function. ij is the true sample label, which is a one-hot encoded vector, where y ij =1 means sample i belongs to category j, otherwise y ij =0. is the probability value predicted by the model, indicating the probability that sample i belongs to category j. N is the number of samples, and C is the number of categories.

[0064] Step 003: Input the recognition result of the large-scale garbage type recognition model on the large-scale garbage image sample into a preset deep learning model for training to obtain a trained large-scale garbage weight prediction model.

[0065] The large-sized garbage weight prediction model is used to predict the weight information of large-sized garbage. The large-sized garbage weight prediction model can determine the material and size information of the large-sized garbage based on the recognition results of the aforementioned large-sized garbage type recognition model, and predict its weight by inputting this information into the large-sized garbage weight prediction model. Since large-sized garbage may not be made of the same material, in an embodiment of the present invention, the large-sized garbage name is predicted, and the general material structure, size and corresponding weight of the same type of large-sized garbage are used for training, thereby obtaining a large-sized garbage weight prediction model. The specific training process of the large-sized garbage weight prediction model can refer to the training method of the aforementioned large-sized garbage type recognition model, which will not be repeated here.

[0066] Among them, after training to obtain a large-scale garbage type recognition model and a large-scale garbage weight prediction model, an embodiment of the present invention inputs the large-scale garbage image into the large-scale garbage type recognition model in real time to identify the large-scale garbage type and large-scale garbage size information of each large-scale garbage in the large-scale garbage image; inputs the large-scale garbage type and large-scale garbage size information of each large-scale garbage into the large-scale garbage weight prediction model for prediction to obtain the weight information of each large-scale garbage; and counts the number of large-scale garbage in each large-scale garbage image, the corresponding large-scale garbage type, large-scale garbage size information and weight information in real time, thereby obtaining the large-scale garbage data of the large-scale garbage recycling site.

[0067] The embodiment of the present invention sets an estimation model for the type and weight of large garbage, so that the quantity and type of large garbage at the large garbage recycling site can be determined in real time. After obtaining the large garbage data at the large garbage recycling site, the large garbage data at the large garbage recycling site is remotely sent to the terminal of the staff, so that the staff can remotely understand the throwing situation of large garbage in real time and clearly even if they are not at the large garbage recycling site.

[0068] Step 130: A loading model is used to predict the large-scale garbage data and vehicle information, and the number of vehicles to be loaded and the large-scale garbage loading plan corresponding to each vehicle to be loaded are determined, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage.

[0069] Among them, the loading model is obtained by training the deep reinforcement learning algorithm using the objective function and constraints according to the large garbage data samples and vehicle information; the objective function is used to maximize the space utilization of the vehicle; the constraints include garbage type constraints, size constraints, weight constraints of large garbage and space constraints of the vehicle.

[0070] Specifically, the training process of the loading model is as follows:

[0071] Step 010: Define the objective function and constraints. The objective function is used to maximize the space utilization of the vehicle; the constraints include the type constraint, size constraint, weight constraint of large garbage and the space constraint of the vehicle.

[0072] Step 011: Obtain large garbage loading samples, obtain the size, weight and other data of the large garbage samples, and the corresponding vehicle information, such as the size and space information of the compartment. Among them, the large garbage samples are obtained by using the existing large furniture and supplies information. The embodiment of the present invention also generates more samples through data enhancement technology, such as rotation, flipping, scaling and other transformations.

[0073] Step 012: Use a deep learning framework to build a neural network model. The model may contain multiple hidden layers, each of which uses an activation function such as ReLU. The deep learning framework can be TensorFlow or Keras algorithm. According to the objective function and constraints, the neural network model is trained using large garbage samples through an optimizer and loss function to obtain a trained loading model.

[0074] In the embodiment of the present invention, the large-sized garbage loading plan includes the relative positions of the large-sized garbage to be loaded in the same vehicle to be loaded.

[0075] Among them, since the complete and undeformed large-sized garbage can generally be directly recycled as second-hand, the complete and undeformed large-sized garbage can be loaded together and transported in a centralized manner. Therefore, a loading plan can be provided separately for the complete and undeformed large-sized garbage. Specifically, the loading model is used to predict the number of vehicles to be loaded with the complete and undeformed large-sized garbage and the vehicle information, and the first large-sized garbage loading plan is determined. According to the first large-sized garbage loading plan, the complete and undeformed large-sized garbage is loaded on the same vehicle to be loaded; the first large-sized garbage loading plan includes the image information of the large-sized garbage in the image and the position indication of the large-sized garbage loaded on the vehicle to be loaded.

[0076] In the embodiment of the present invention, in addition to the large pieces of garbage that are intact and not deformed, the loading model is used to predict the number of vehicles to be loaded and the second large piece of garbage loading plan based on the damaged large pieces of metal material, the deformed large pieces of metal material, the damaged large pieces of fabric material, the deformed large pieces of fabric material, the damaged large pieces of wood material, the deformed large pieces of wood material, the damaged large pieces of wood material, the deformed large pieces of wood material and the vehicle information. Among them, in the second large piece of garbage loading plan, large pieces of garbage of the same material are preferentially allocated to the same vehicle to be loaded. Afterwards, according to the second large piece of garbage loading plan, the damaged large pieces of metal material, the deformed large pieces of metal material, the damaged large pieces of fabric material, the deformed large pieces of fabric material, the damaged large pieces of wood material, the deformed large pieces of wood material, the damaged large pieces of wood material, the deformed large pieces of wood material, the damaged large pieces of wood material, the deformed large pieces of wood material and the deformed large pieces of wood are loaded onto the corresponding vehicles to be loaded.

[0077] Wherein, at least one illegal loading camera device is also arranged above the loading space of the vehicle to be loaded; when loading the large garbage into the vehicle to be loaded, the method further comprises:

[0078] The illegal loading camera device acquires loading images during the loading process in real time;

[0079] According to the loading image, it is determined whether there is any non-compliant garbage on the vehicle to be loaded; wherein the non-compliant garbage at least includes: whether other garbage other than large garbage is illegally loaded. In this way, the staff can check in real time whether the driver of the vehicle to be loaded has mistakenly loaded or illegally loaded other garbage other than large garbage.

[0080] In an embodiment of the present invention, after loading large pieces of garbage, a trigger operation is obtained according to the user's path planning, and the information of each large piece of garbage sorting and processing point and the information of the second-hand recycling point within the target range are also obtained; the information of the large piece of garbage sorting and processing point includes the location information and current working status information of the large piece of garbage sorting and processing point; according to the information of the second-hand recycling point, the first path planning information is generated for the vehicle loaded with complete and undeformed large pieces of garbage; according to the information of each large piece of garbage sorting and processing point, the large piece of garbage sorting and processing point corresponding to the remaining loaded vehicles is determined, and the second path planning information of the path is generated respectively. Among them, the information of each large piece of garbage sorting and processing point is the address, current processing volume and other information of each large piece of garbage sorting and processing point uploaded by each large piece of garbage sorting and processing point or obtained in real time according to the second intelligent video unit. In this way, the transportation point of large pieces of garbage can be effectively adjusted in real time according to the situation of each large piece of garbage sorting and processing point. At the same time, through path planning, the optimal route can be provided, so that the transportation efficiency can be improved.

[0081] Step 140: During the sorting process at the large-scale garbage sorting and processing point, the large-scale garbage sorting video is obtained in real time through the second intelligent monitoring unit, and the large-scale garbage sorting video is identified and detected to determine whether there is a sorting error in the large-scale garbage sorting process. When a sorting error occurs, a sorting error prompt is sent to the user in real time.

[0082] Among them, the large-scale garbage sorting and processing point classifies and disassembles the received large-scale garbage. For mixed large-scale garbage, electric tools are used to separate metal frames, wooden parts, disassemble mixed garbage, and classify and place them in different areas for subsequent different crushing treatments. In order to improve the processing efficiency and resource recycling rate in the subsequent crushing process, the probability of misclassification of different types of garbage should be minimized. Therefore, the embodiment of the present invention inputs the large-scale garbage sorting video into the garbage sorting detection model to determine whether there are large-scale garbage with sorting errors in each sorting area; wherein the garbage sorting detection model is pre-trained according to the sorting video sample input into the neural network model, and the sorting video sample includes the sorting sample video and the sorting garbage type labels of each sorting area in the large-scale garbage sorting processing point in the sorting sample video; the sorting garbage labels include metal parts, wooden parts, and disassembled mixed garbage.

[0083] After the disassembly and classification are completed, different types of disassembled waste are placed in different resource processing tools. For example, metal waste is uniformly crushed and then extracted and classified to obtain different metal materials. For large non-metallic waste, the large waste is crushed into small pieces, and then the organic components in the waste are decomposed by microorganisms, and the indecomposable waste is separated, and finally the usable waste compost is selected to achieve the purpose of recycling waste resources.

[0084] The embodiment of the present invention determines the large-scale garbage data such as the quantity, type and basic attribute information of the large-scale garbage by setting up multiple video monitoring devices at the large-scale garbage recycling point and each large-scale garbage sorting and processing point through real-time analysis and statistics of the large-scale garbage images, and determines the required number of vehicles and the loading plan of each vehicle to be loaded based on these large-scale garbage data and vehicle information, so that the large-scale garbage can be pre-classified before being loaded into the vehicle, and the number of vehicles and loading space can be reasonably used. Furthermore, sorting errors can be identified by a sorting detection model at each large-scale garbage sorting and processing point, which can improve resource utilization. Through the above scheme, the processing efficiency of large-scale garbage can be effectively improved and the transportation cost can be reduced.

[0085] Figure 3 FIG. 1 is a schematic diagram showing the structure of a large-sized garbage treatment system provided by an embodiment of the present invention. Figure 3 As shown, the system 200 includes a video monitoring device. The video monitoring device includes at least one first intelligent monitoring unit 210 disposed at a large-sized garbage collection location, at least one second intelligent monitoring unit 240 disposed at a large-sized garbage sorting and processing location, an analysis and statistics unit 220, a loading prediction unit 230, and a sorting detection unit 250.

[0086] The first intelligent monitoring unit 210 is used to obtain in real time a large-scale garbage image corresponding to the large-scale garbage accumulation at the large-scale garbage collection site;

[0087] The analysis and statistics unit 220 is used to analyze and count the large-scale garbage data of the large-scale garbage collection site in real time according to the large-scale garbage image; the large-scale garbage data includes the number of large-scale garbage, the type of large-scale garbage and the basic attribute information of large-scale garbage; the basic attribute information of large-scale garbage includes the size information and weight information of large-scale garbage;

[0088] The loading prediction unit 230 is used to predict using a loading model based on the large-scale garbage data and vehicle information, determine the number of vehicles to be loaded and the large-scale garbage loading plan corresponding to each vehicle to be loaded, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage; the loading model is obtained by training a deep reinforcement learning algorithm based on the large-scale garbage data samples and vehicle information using an objective function and constraints; the objective function is used to maximize the space utilization of the vehicle; the constraints include the garbage type constraint, size constraint, weight constraint of the large-scale garbage and the space constraint of the vehicle; the large-scale garbage loading plan includes the relative position of each large-scale garbage to be loaded in the same vehicle to be loaded;

[0089] The sorting detection unit 250 is used to obtain the large-piece garbage sorting video in real time through the second intelligent monitoring unit during the sorting process at the large-piece garbage sorting processing point, and identify and detect the large-piece garbage sorting video to determine whether there is a sorting error in the large-piece garbage sorting process. When a sorting error occurs, a sorting error prompt is sent to the user in real time.

[0090] In an optional manner, the system further includes:

[0091] An image sample acquisition unit is used to acquire a large-piece garbage image sample; the large-piece garbage image sample includes a large-piece garbage sample image and a corresponding sample label; the sample label includes a large-piece garbage type label, a large-piece garbage size label and a weight label corresponding to the large-piece garbage sample image;

[0092] A large-scale garbage type recognition model training unit is used to input the large-scale garbage image sample into a preset YOLOV5 model for training to obtain a trained large-scale garbage type recognition model; the large-scale garbage type recognition model is used to identify large-scale garbage, large-scale garbage type and large-scale garbage size in the large-scale garbage image;

[0093] The large-scale garbage weight prediction model training unit is used to input the recognition result of the large-scale garbage type recognition model on the large-scale garbage image sample into a preset deep learning model for training to obtain a trained large-scale garbage weight prediction model; the large-scale garbage weight prediction model is used to predict the weight information of large-scale garbage.

[0094] In an optional manner, the real-time analysis and statistics of the bulky waste data at the bulky waste collection site based on the bulky waste image includes:

[0095] By inputting the large-scale garbage image into a large-scale garbage type recognition model in real time, the large-scale garbage type and large-scale garbage size information of each large-scale garbage in the large-scale garbage image are identified;

[0096] Inputting the large-piece garbage type and large-piece garbage size information of each large-piece garbage into the large-piece garbage weight prediction model for prediction to obtain the weight information of each large-piece garbage;

[0097] The number of large-sized garbage in each of the large-sized garbage images, the corresponding large-sized garbage type, the large-sized garbage size information and the weight information are counted in real time, so as to obtain the large-sized garbage data of the large-sized garbage recycling site.

[0098] In an optional manner, the types of large garbage include at least one or more of complete and undeformed large garbage, damaged metal large garbage, deformed metal large garbage, damaged fabric large garbage, deformed fabric large garbage, damaged wooden large garbage, deformed wooden large garbage, damaged wooden large garbage, and deformed wooden large garbage.

[0099] In an optional manner, the prediction is performed using a loading model based on the large-scale garbage data and vehicle information to determine the number of vehicles to be loaded and the large-scale garbage information corresponding to each vehicle to be loaded, so that the user calls the vehicle to be loaded to load the corresponding large-scale garbage, including:

[0100] According to the complete and undeformed large-sized garbage and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded with the complete and undeformed large-sized garbage and the first large-sized garbage loading plan;

[0101] According to the damaged large-sized metal garbage, deformed large-sized metal garbage, damaged large-sized fabric garbage, deformed large-sized fabric garbage, damaged large-sized wooden garbage, deformed large-sized wooden garbage, damaged large-sized wooden garbage, deformed large-sized wooden garbage and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded and the loading plan for the second largest garbage;

[0102] According to the first large-piece garbage loading plan, the intact and undeformed large-piece garbage is loaded on the same vehicle to be loaded; the first large-piece garbage loading plan includes image information of the large-piece garbage in the image and a position indication of the large-piece garbage loaded on the vehicle to be loaded;

[0103] According to the second large-piece garbage loading plan, the damaged large-piece metal garbage, deformed large-piece metal garbage, damaged large-piece fabric garbage, deformed large-piece fabric garbage, damaged large-piece wooden garbage, deformed large-piece wooden garbage, damaged large-piece wooden garbage, and deformed large-piece wooden garbage are respectively loaded onto the corresponding vehicles to be loaded.

[0104] In an optional manner, the system further comprises at least one illegal loading camera device and an illegal loading detection unit arranged above the loading space of the vehicle to be loaded:

[0105] When the device is in the process of loading large pieces of garbage into the vehicle to be loaded, the illegal loading camera device is used to obtain the loading image during the loading process in real time;

[0106] The violation detection unit is used to determine whether there is non-compliant garbage on the vehicle to be loaded according to the loading image; wherein the non-compliant garbage at least includes: whether other garbage except large garbage is loaded illegally.

[0107] In an optional manner, the system further includes:

[0108] A site information acquisition unit is used to acquire information about each large-scale waste sorting and processing point and information about second-hand recycling points within a target range; the information about the large-scale waste sorting and processing point includes location information and current working status information of the large-scale waste sorting and processing point;

[0109] A path planning unit, configured to generate first path planning information for a vehicle loaded with intact and undeformed large garbage according to the information of the second-hand recycling point;

[0110] The path planning unit is also used to determine the large-piece waste sorting and processing points corresponding to the remaining loading vehicles according to the information of each large-piece waste sorting and processing point, and generate the second path planning information of the path respectively.

[0111] In an optional manner, during the sorting process at the large-scale waste sorting processing point, the large-scale waste sorting video is obtained in real time by the second intelligent monitoring unit, and the large-scale waste sorting video is identified and detected to determine whether there is a sorting error in the large-scale waste sorting process. When a sorting error occurs, a sorting error prompt is sent to the user in real time, including:

[0112] The large-piece garbage sorting video is input into the garbage sorting detection model to determine whether there is large-piece garbage that is sorted incorrectly in each sorting area; wherein the garbage sorting detection model is pre-trained based on the sorting video samples input into the neural network model, and the sorting video samples include the sorting sample videos and the sorting garbage type labels of each sorting area in the large-piece garbage sorting processing point in the sorting sample videos; the sorting garbage labels include metal parts, wooden parts, and disassembled mixed garbage.

[0113] The embodiment of the present invention determines the large-scale garbage data such as the quantity, type and basic attribute information of the large-scale garbage by setting up multiple video monitoring devices at the large-scale garbage recycling point and each large-scale garbage sorting and processing point through real-time analysis and statistics of the large-scale garbage images, and determines the required number of vehicles and the loading plan of each vehicle to be loaded based on these large-scale garbage data and vehicle information, so that the large-scale garbage can be pre-classified before being loaded into the vehicle, and the number of vehicles and loading space can be reasonably used. Furthermore, sorting errors can be identified by a sorting detection model at each large-scale garbage sorting and processing point, which can improve resource utilization. Through the above scheme, the processing efficiency of large-scale garbage can be effectively improved and the transportation cost can be reduced.

[0114] The algorithm or display provided herein is not inherently related to any particular computer, virtual system or other equipment. Various general purpose systems can also be used together with the teachings based on this. According to the above description, it is obvious to construct the structure required for this type of system. In addition, the embodiment of the present invention is not directed to any specific programming language yet. It should be understood that various programming languages ​​can be utilized to realize the content of the present invention described herein, and the description made to specific languages ​​above is for disclosing the best mode of the present invention.

[0115] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. In some instances, well-known methods, structures and techniques are not shown in detail so as not to obscure the understanding of this description.

[0116] Similarly, it should be understood that in order to streamline the present invention and aid in understanding one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. However, this method of disclosure should not be interpreted as reflecting an intention that the claimed invention requires more features than those expressly recited in each claim.

[0117] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive, all features disclosed in this specification (including the accompanying claims, abstracts and drawings) and all processes or units of any method or device disclosed in this manner may be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstracts and drawings) may be replaced by an alternative feature providing the same, equivalent or similar purpose.

[0118] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A method for processing large-sized garbage, characterized in that: The method is based on a large-scale garbage processing system, which includes a video monitoring device; the video monitoring device includes at least one first intelligent monitoring unit arranged at a large-scale garbage collection location and at least one second intelligent monitoring unit arranged at a large-scale garbage sorting and processing point, and the method includes: Acquire, in real time, through the first intelligent monitoring unit, a large-scale garbage image corresponding to the large-scale accumulated garbage at the large-scale garbage collection site; According to the large-scale garbage image, the large-scale garbage data of the large-scale garbage collection site is analyzed and counted in real time; the large-scale garbage data includes the number of large-scale garbage, the type of large-scale garbage and the basic attribute information of large-scale garbage; the basic attribute information of large-scale garbage includes the size information and weight information of large-scale garbage; According to the large-scale garbage data and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded and the large-scale garbage loading plan corresponding to each vehicle to be loaded, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage; the loading model is obtained by training a deep reinforcement learning algorithm using an objective function and constraints based on the large-scale garbage data samples and vehicle information; the objective function is used to maximize the space utilization of the vehicle; the constraints include the garbage type constraints, size constraints, weight constraints of the large-scale garbage and the space constraints of the vehicle; the large-scale garbage loading plan includes the relative positions of each large-scale garbage to be loaded in the same vehicle to be loaded; During the sorting process at the large-scale garbage sorting and processing point, the large-scale garbage sorting video is obtained in real time through the second intelligent monitoring unit, and the large-scale garbage sorting video is identified and detected to determine whether there is a sorting error in the large-scale garbage sorting process. When a sorting error occurs, a sorting error prompt is sent to the user in real time.

2. The method according to claim 1, characterized in that Before analyzing and counting the large-scale garbage data at the large-scale garbage collection site in real time according to the large-scale garbage image, the method further comprises: Obtaining a large garbage image sample; the large garbage image sample includes a large garbage sample image and a corresponding sample label; the sample label includes a large garbage type label, a large garbage size label, and a weight label corresponding to the large garbage sample image; Input the large garbage image sample into a preset YOLOV5 model for training to obtain a trained large garbage type recognition model; the large garbage type recognition model is used to identify large garbage, large garbage type and large garbage size in the large garbage image; The recognition result of the large-scale garbage type recognition model on the large-scale garbage image sample is input into a preset deep learning model for training to obtain a trained large-scale garbage weight prediction model; the large-scale garbage weight prediction model is used to predict the weight information of large-scale garbage.

3. The method according to claim 2, characterized in that The step of analyzing and counting the large-scale garbage data at the large-scale garbage collection site in real time according to the large-scale garbage image comprises: By inputting the large-scale garbage image into a large-scale garbage type recognition model in real time, the large-scale garbage type and large-scale garbage size information of each large-scale garbage in the large-scale garbage image are identified; Inputting the large-piece garbage type and large-piece garbage size information of each large-piece garbage into the large-piece garbage weight prediction model for prediction to obtain the weight information of each large-piece garbage; The number of large-sized garbage in each of the large-sized garbage images, the corresponding large-sized garbage type, the large-sized garbage size information and the weight information are counted in real time, so as to obtain the large-sized garbage data of the large-sized garbage recycling site.

4. The method according to claim 2, characterized in that: The types of large garbage include at least one or more of complete and undeformed large garbage, damaged metal large garbage, deformed metal large garbage, damaged fabric large garbage, deformed fabric large garbage, damaged wooden large garbage, deformed wooden large garbage, damaged wooden large garbage, and deformed wooden large garbage.

5. The method according to claim 4, characterized in that The method of using a loading model to predict based on the large-scale garbage data and vehicle information, determining the number of vehicles to be loaded and the large-scale garbage information corresponding to each vehicle to be loaded, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage, includes: According to the complete and undeformed large-sized garbage and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded with the complete and undeformed large-sized garbage and the first large-sized garbage loading plan; According to the damaged large-sized metal garbage, deformed large-sized metal garbage, damaged large-sized fabric garbage, deformed large-sized fabric garbage, damaged large-sized wooden garbage, deformed large-sized wooden garbage, damaged large-sized wooden garbage, deformed large-sized wooden garbage and vehicle information, a loading model is used to predict and determine the number of vehicles to be loaded and the loading plan for the second largest garbage; According to the first large-piece garbage loading plan, the intact and undeformed large-piece garbage is loaded on the same vehicle to be loaded; the first large-piece garbage loading plan includes image information of the large-piece garbage in the image and a position indication of the large-piece garbage loaded on the vehicle to be loaded; According to the second large-piece garbage loading plan, the damaged large-piece metal garbage, deformed large-piece metal garbage, damaged large-piece fabric garbage, deformed large-piece fabric garbage, damaged large-piece wooden garbage, deformed large-piece wooden garbage, damaged large-piece wooden garbage, and deformed large-piece wooden garbage are respectively loaded onto the corresponding vehicles to be loaded.

6. The method according to claim 5, characterized in that At least one illegal loading camera is also arranged above the loading space of the vehicle to be loaded; when loading the large garbage into the vehicle to be loaded, the method further comprises: The illegal loading camera device acquires loading images during the loading process in real time; Determine whether there is any non-compliant garbage on the vehicle to be loaded according to the loading image; wherein the non-compliant garbage at least includes: whether other garbage except large garbage is illegally loaded.

7. The method according to claim 5, characterized in that In the sorting process at the large-scale waste sorting processing point, the large-scale waste sorting video is obtained in real time by the second intelligent monitoring unit, and the large-scale waste sorting video is identified and detected to determine whether there is a sorting error in the large-scale waste sorting process. When there is a sorting error, before sending a sorting error prompt to the user in real time, the method further includes: Obtaining information on each large-scale waste sorting and processing point and information on second-hand recycling points within the target range; the information on the large-scale waste sorting and processing point includes location information and current working status information of the large-scale waste sorting and processing point; Generating first path planning information for a vehicle loaded with intact and undeformed large garbage based on the information of the second-hand recycling point; According to the information of each large-scale garbage sorting and processing point, the large-scale garbage sorting and processing points corresponding to the remaining loading vehicles are determined, and the second path planning information of the path is generated respectively.

8. The method according to any one of claims 1 to 7, characterized in that: In the sorting process at the large-scale garbage sorting processing point, the large-scale garbage sorting video is obtained in real time through the second intelligent monitoring unit, and the large-scale garbage sorting video is identified and detected to determine whether there is a sorting error in the large-scale garbage sorting process. When there is a sorting error, a sorting error prompt is sent to the user in real time, including: The large-piece garbage sorting video is input into the garbage sorting detection model to determine whether there is large-piece garbage that is sorted incorrectly in each sorting area; wherein the garbage sorting detection model is pre-trained based on the sorting video samples input into the neural network model, and the sorting video samples include the sorting sample videos and the sorting garbage type labels of each sorting area in the large-piece garbage sorting processing point in the sorting sample videos; the sorting garbage labels include metal parts, wooden parts, and disassembled mixed garbage.

9. A bulky garbage disposal system, characterized in that: The system includes a video monitoring device; the video monitoring device includes at least one first intelligent monitoring unit arranged at a large-sized garbage collection location, at least one second intelligent monitoring unit arranged at a large-sized garbage sorting and processing location, an analysis and statistics unit, a loading prediction unit and a sorting detection unit; wherein: The first intelligent monitoring unit is used to obtain in real time the large-scale garbage images corresponding to the large-scale garbage accumulation at the large-scale garbage collection site; An analysis and statistics unit is used to analyze and count the large-scale garbage data of the large-scale garbage collection site in real time according to the large-scale garbage image; the large-scale garbage data includes the number of large-scale garbage, the type of large-scale garbage and the basic attribute information of large-scale garbage; the basic attribute information of large-scale garbage includes the size information and weight information of large-scale garbage; A loading prediction unit is used to predict using a loading model based on the large-scale garbage data and vehicle information, determine the number of vehicles to be loaded and the large-scale garbage loading plan corresponding to each vehicle to be loaded, so that the user can call the vehicle to be loaded to load the corresponding large-scale garbage; the loading model is obtained by training a deep reinforcement learning algorithm based on the large-scale garbage data samples and vehicle information using an objective function and constraints; the objective function is used to maximize the space utilization of the vehicle; the constraints include the garbage type constraint, size constraint, weight constraint of the large-scale garbage and the space constraint of the vehicle; the large-scale garbage loading plan includes the relative position of each large-scale garbage to be loaded in the same vehicle to be loaded; The sorting detection unit is used to obtain the large-piece garbage sorting video in real time through the second intelligent monitoring unit during the sorting process at the large-piece garbage sorting processing point, and identify and detect the large-piece garbage sorting video to determine whether there is a sorting error in the large-piece garbage sorting process, and when there is a sorting error, send a sorting error prompt to the user in real time.

10. The system according to claim 9, characterized in that The system further comprises: An image sample acquisition unit is used to acquire a large-piece garbage image sample; the large-piece garbage image sample includes a large-piece garbage sample image and a corresponding sample label; the sample label includes a large-piece garbage type label, a large-piece garbage size label and a weight label corresponding to the large-piece garbage sample image; A large-scale garbage type recognition model training unit is used to input the large-scale garbage image sample into a preset YOLOV5 model for training to obtain a trained large-scale garbage type recognition model; the large-scale garbage type recognition model is used to identify large-scale garbage, large-scale garbage type and large-scale garbage size in the large-scale garbage image; The large-scale garbage weight prediction model training unit is used to input the recognition result of the large-scale garbage type recognition model on the large-scale garbage image sample into a preset deep learning model for training to obtain a trained large-scale garbage weight prediction model; the large-scale garbage weight prediction model is used to predict the weight information of large-scale garbage.

Images