Intelligent management method and device for mixed arrangement of people and vehicles

Abstract

The invention discloses an intelligent management method and device for mixed arrangement of people and vehicles, and relates to the technical field of intelligent scheduling. Acquiring an environment image in the target area, and performing road recognition on the environment image to obtain a target road image; performing garbage detection on the target road image to obtain garbage data, and determining delivery vehicle data according to the garbage data; searching a preset database according to the junk data and the delivery vehicle data to obtain a first scheme sorting set, and determining delivery personnel data according to the sorting of each scheme in the first scheme sorting set; and determining a delivery scheme according to the delivery vehicle data and the delivery personnel data. By performing garbage detection on the target road image and generating garbage data, the system can determine the number and type of sanitation vehicles needing to be delivered, and further determine the number and task of delivery personnel according to the data of the delivery vehicles, so that reasonable allocation of resources can be ensured, allocation of the resources is optimized, and the efficiency of the system is improved. The problem of excessive delivery or insufficient delivery is avoided, and the cleaning efficiency is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent scheduling, and specifically relates to an intelligent management method and device for mixed scheduling of people and vehicles. Background Art

[0002] With the acceleration of urbanization, the importance of urban cleaning in ensuring the quality of life of citizens, improving the image of the city and creating a livable environment has become increasingly prominent. In order to ensure the cleanliness and beauty of the city, cleaners and sanitation vehicles have formed an efficient cooperation model and become an important part of modern urban cleaning. Cleaners are mainly responsible for daily street cleaning, garbage classification guidance, maintenance of public facilities and regular detailed cleaning, while sanitation vehicles are responsible for large-scale garbage removal, road dust reduction, road cleaning and other tasks. Sanitation vehicles have efficient operating capabilities in cleaning large areas and large-scale garbage transfer, while cleaners play an irreplaceable role in detailed cleaning, garbage classification and maintenance of local areas.

[0003] In current urban sanitation operations, technicians usually rely on analysis based on taken pictures and historical experience to determine whether to dispatch cleaning vehicles and personnel. Due to the lack of accurate judgment of the actual environmental conditions, misjudgment by technicians may lead to excessive dispatch of cleaning resources, resulting in waste, or insufficient dispatch, resulting in low cleaning efficiency. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that due to the lack of accurate judgment of the actual environmental situation, the misjudgment of technicians may lead to excessive dispatch of cleaning resources, resulting in waste, or insufficient dispatch, resulting in low cleaning efficiency, and to propose an intelligent management method and device for mixed human and vehicle arrangements.

[0005] In a first aspect of the present invention, a smart management method for mixed crowding of people and vehicles is first proposed, the method comprising:

[0006] Collecting an environmental image in a target area, and performing road recognition on the environmental image to obtain a target road image;

[0007] Performing garbage detection on the target road image to obtain garbage data, and determining dispatch vehicle data according to the garbage data;

[0008] Searching a preset database according to the garbage data and the delivery vehicle data to obtain a first solution sorting set, and determining the delivery personnel data according to the sorting of each solution in the first solution sorting set;

[0009] A delivery plan is determined based on the delivery vehicle data and the delivery personnel data.

[0010] Optionally, performing road recognition on the environment image to obtain a target road image includes:

[0011] Preprocessing the environment image to obtain an initial image, performing grayscale conversion on the initial image to obtain a target grayscale image, substituting the target grayscale image into a preset encoder to perform feature extraction to obtain a first feature map;

[0012] Processing the first feature map by sub-pixel convolution to obtain a second feature map, and concatenating the first feature map and the second feature map to obtain a target feature map;

[0013] The target feature map is processed by a segmentation algorithm to obtain a road distribution area, and the environment image is cut according to the road distribution area to obtain an image containing the road distribution area, which is recorded as a target road image.

[0014] Optionally, performing garbage detection on the target road image to obtain garbage data includes:

[0015] Performing noise filtering and contrast enhancement on the target road image to obtain a first road image, and converting the first road image into an HSV color space to obtain a second road image;

[0016] Substituting the second road image into a preset garbage detection model to obtain bounding boxes and classification results corresponding to a plurality of dry garbage;

[0017] Performing local histogram equalization on the first road image and then applying high-pass filtering to obtain a third road image;

[0018] Substituting the third road image into a preset segmentation network model to perform pixel segmentation to obtain an initial mask, and performing morphological processing on the initial mask to obtain a target wet garbage area;

[0019] Segmenting the third road image according to the target wet garbage area to obtain a fourth road image, and substituting the fourth road image into a preset deep learning classification model to obtain a type corresponding to each wet garbage area;

[0020] The bounding box and classification result corresponding to each dry garbage area and the type corresponding to each wet garbage area are recorded as garbage data.

[0021] Optionally, determining the delivery vehicle data according to the garbage data includes:

[0022] Determine the target total amount of dry garbage based on the bounding boxes and classification results corresponding to all dry garbage, and determine the target number of dry garbage trucks to be dispatched based on the target total amount of dry garbage;

[0023] The total area of ​​the target wet garbage area is determined according to each wet garbage area, and the target wet garbage truck dispatch quantity corresponding to the total area of ​​the target wet garbage area in the historical data is obtained according to the total area of ​​the target wet garbage area.

[0024] Optionally, searching a preset database to obtain a first solution sorting set according to the garbage data and the delivery vehicle data, and determining the delivery personnel data according to the sorting of each solution in the first solution sorting set includes:

[0025] According to the target total amount of dry garbage, the target total area of ​​wet garbage area and the target number of wet garbage trucks to be delivered, a first solution sorting set is obtained by searching for a solution that satisfies the target in the preset database;

[0026] According to the total area of ​​the target wet garbage area, the target total amount of dry garbage, and the target number of dry garbage trucks to be delivered, a second solution ranking set is obtained by searching for a solution that satisfies the requirements in the preset database;

[0027] Obtain the intersection of the first solution sorting set and the second solution sorting set to obtain a target solution set. If the target solution set is not an empty set, obtain the solution with the least number of dispatched personnel in the target solution set and record it as a preset solution.

[0028] Calculate the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset plan and the target total amount of dry garbage and the target total area of ​​wet garbage. If the similarity is greater than a preset threshold, the number of people delivered in the preset plan is recorded as delivery personnel data.

[0029] In a second aspect of the present invention, a smart management device for mixed crowding of people and vehicles is provided, comprising:

[0030] A road recognition module is used to collect an environmental image in a target area and perform road recognition on the environmental image to obtain a target road image;

[0031] A garbage detection module, used for performing garbage detection on the target road image to obtain garbage data, and determining the dispatch vehicle data according to the garbage data;

[0032] A personnel dispatch determination module, configured to search a preset database according to the garbage data and the dispatch vehicle data to obtain a first scheme sorting set, and determine the dispatch personnel data according to the sorting of each scheme in the first scheme sorting set;

[0033] The delivery plan determination module is used to determine the delivery plan based on the delivery vehicle data and the delivery personnel data.

[0034] Optionally, the road recognition module includes:

[0035] A road recognition module, used for preprocessing the environment image to obtain an initial image, performing grayscale conversion on the initial image to obtain a target grayscale image, and substituting the target grayscale image into a preset encoder to perform feature extraction to obtain a first feature map;

[0036] An image stitching module, configured to process the first feature map by sub-pixel convolution to obtain a second feature map, and stitch the first feature map and the second feature map to obtain a target feature map;

[0037] The road recognition module is used to process the target feature map through a segmentation algorithm to obtain a road distribution area, and cut the environment image according to the road distribution area to obtain an image containing the road distribution area and record it as a target road image.

[0038] Optionally, the garbage detection module includes:

[0039] An image enhancement module, configured to perform noise filtering and contrast enhancement on the target road image to obtain a first road image, and convert the first road image into an HSV color space to obtain a second road image;

[0040] A dry garbage detection module, used to substitute the second road image into a preset garbage detection model to obtain bounding boxes and classification results corresponding to a plurality of dry garbage;

[0041] An image filtering module, configured to perform local histogram equalization on the first road image and then apply high-pass filtering to obtain a third road image;

[0042] A wet garbage area determination module is used to substitute the third road image into a preset segmentation network model to perform pixel segmentation to obtain an initial mask, and perform morphological processing on the initial mask to obtain a target wet garbage area;

[0043] A wet garbage detection module, configured to segment the third road image according to the target wet garbage area to obtain a fourth road image, and substitute the fourth road image into a preset deep learning classification model to obtain a type corresponding to each wet garbage area;

[0044] The garbage data determination module is used to record the border and classification result corresponding to each dry garbage and the type corresponding to each wet garbage area as garbage data.

[0045] Optionally, the garbage detection module further includes:

[0046] A dry garbage truck quantity determination module is used to determine the target total amount of dry garbage according to the bounding boxes and classification results corresponding to all dry garbage, and determine the target number of dry garbage trucks to be dispatched according to the target total amount of dry garbage;

[0047] The wet garbage truck quantity determination module is used to determine the total area of ​​the target wet garbage area according to each wet garbage area, and obtain the target wet garbage truck dispatch quantity corresponding to the total area of ​​the target wet garbage area in the historical data according to the total area of ​​the target wet garbage area.

[0048] Optionally, the personnel dispatch determination module includes:

[0049] A first scheme ranking set determination module is used to find a scheme that satisfies the target total amount of dry garbage, the target total area of ​​wet garbage area and the target number of wet garbage trucks to be delivered to obtain a first scheme ranking set;

[0050] A second scheme ranking set determination module is used to find a scheme that satisfies the requirements in a preset database according to the total area of ​​the target wet garbage area, the target total amount of dry garbage, and the target number of dry garbage trucks to be delivered, to obtain a second scheme ranking set;

[0051] A preset solution determination module is used to obtain the intersection of the first solution sorting set and the second solution sorting set to obtain a target solution set. If the target solution set is not an empty set, the solution with the least number of dispatched personnel in the target solution set is obtained and recorded as the preset solution.

[0052] The similarity calculation module is used to calculate the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset plan and the target total amount of dry garbage and the total area of ​​wet garbage. If the similarity is greater than a preset threshold, the number of people delivered in the preset plan is recorded as delivery personnel data.

[0053] Beneficial effects of the present invention:

[0054] The present invention proposes an intelligent management method for mixed human and vehicle arrangements, which collects environmental images in the target area, performs road recognition on the environmental images to obtain target road images; performs garbage detection on the target road images to obtain garbage data, and determines the delivery vehicle data based on the garbage data; searches a preset database based on the garbage data and the delivery vehicle data to obtain a first scheme sorting set, and determines the delivery personnel data based on the sorting of each scheme in the first scheme sorting set; and determines the delivery scheme based on the delivery vehicle data and the delivery personnel data. By performing garbage detection on the target road image and generating garbage data, the degree of dirtiness and types of garbage in the area can be accurately assessed, the system can determine the number and type of sanitation vehicles required for delivery, and further determine the number and tasks of delivery personnel based on the delivery vehicle data, which can ensure the rational allocation of resources, thereby optimizing the allocation of resources, avoiding the problem of over-delivery or under-delivery, and improving cleaning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The present invention will be further described below in conjunction with the accompanying drawings.

[0056] Figure 1A flowchart of an intelligent management method for mixed crowding of people and vehicles is provided for an embodiment of the present invention;

[0057] Figure 2 A structural schematic diagram of an intelligent management device for mixed crowds of people and vehicles is provided for an embodiment of the present invention. DETAILED DESCRIPTION

[0058] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there can be three relationships. For example, A and B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, in the present invention, the description of "first", "second", etc. is only used for descriptive purposes, and cannot be understood as indicating or implying its relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" can explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the ability of ordinary technicians in the field to implement. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0059] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0060] The embodiment of the present invention provides an intelligent management method for mixed crowding of people and vehicles. Figure 1 , Figure 1 A flowchart of a method for intelligent management of mixed traffic of people and vehicles provided in an embodiment of the present invention. The method comprises the following steps:

[0061] S101, collecting an environmental image within a target area, and performing road recognition on the environmental image to obtain a target road image;

[0062] S102, performing garbage detection on the target road image to obtain garbage data, and determining the dispatch vehicle data according to the garbage data;

[0063] S103, searching a preset database according to the garbage data and the delivery vehicle data to obtain a first solution sorting set, and determining the delivery personnel data according to the sorting of each solution in the first solution sorting set;

[0064] S104, determining a delivery plan based on the delivery vehicle data and the delivery personnel data.

[0065] Based on an intelligent management method for mixed traffic of people and vehicles provided by an embodiment of the present invention, by performing garbage detection on the target road image and generating garbage data, the degree of dirtiness and type of garbage in the area can be accurately assessed. The system can determine the number and type of sanitation vehicles that need to be delivered, and further determine the number and tasks of delivery personnel based on the delivery vehicle data, thereby ensuring the reasonable allocation of resources, thereby optimizing the allocation of resources, avoiding the problem of over-delivery or under-delivery, and improving cleaning efficiency.

[0066] In one implementation, by collecting environmental images of the target area and performing road recognition, the system can accurately obtain the cleanliness status of the current area in real time, avoid errors in traditional manual judgment, and achieve accurate identification of the environment and real-time assessment of cleaning needs; based on the garbage data obtained from garbage detection, the system can accurately assess the type, quantity and distribution of garbage, thereby accurately determining the sanitation vehicles and personnel that need to be dispatched, avoiding unnecessary waste of resources.

[0067] In one implementation, the target area is a commercial street or alley, the environmental image is taken by a camera installed in the commercial street or alley, and the preset database is used to save the total amount of dry garbage, the total area of ​​wet garbage area, the number of dry garbage trucks delivered, the number of wet garbage trucks delivered, the number of personnel delivered and the corresponding satisfaction levels recorded in the historical treatment plan.

[0068] In one implementation, through garbage data and delivery vehicle data, the system can quickly find the most suitable cleaning plan from a preset database, and determine the optimal number of delivery personnel based on the plan sorting, so that cleaning tasks can be efficiently allocated according to actual needs, avoiding delays and errors caused by human judgment, and ensuring the timeliness and accuracy of task execution.

[0069] In one embodiment, performing road recognition on the environment image to obtain the target road image includes:

[0070] Preprocessing the environment image to obtain an initial image, performing grayscale conversion on the initial image to obtain a target grayscale image, substituting the target grayscale image into a preset encoder to perform feature extraction to obtain a first feature map;

[0071] The first feature map is processed by sub-pixel convolution to obtain a second feature map, and the first feature map and the second feature map are concatenated to obtain a target feature map;

[0072] The target feature map is processed by a segmentation algorithm to obtain a road distribution area, and the environment image is cut according to the road distribution area to obtain an image containing the road distribution area, which is recorded as a target road image.

[0073] In one implementation, preprocessing is the use of existing technologies such as denoising and contrast enhancement to improve the accuracy of subsequent processing and remove irrelevant information, making important elements such as roads and garbage in the target image more prominent; through grayscale conversion and substitution of a preset encoder for feature extraction, the system can extract more concise and effective features from the initial image, which not only reduces the complexity of the data, but also helps the algorithm focus on the most recognizable features, thereby improving the accuracy of image recognition and processing efficiency.

[0074] In one implementation, the preset encoder includes a downsampling module and a convolution module, wherein the downsampling module includes 1 3*3 convolution layer and 2 1*1 convolution layers, and the downsampling module is operated to substitute the input image into the 1*1 convolution layer for convolution and normalize it to obtain a first sampled image, substitute the first sampled image into the 3*3 convolution layer for convolution and normalize it to obtain a second sampled image, perform an addition operation on the first sampled image and the second sampled image, and then substitute them into the 1*1 convolution layer and normalize them to obtain an output image, and the convolution module also includes 1 3*3 convolution layer and 2 1*1 convolution layers, and the convolution module is operated to substitute the input image into the 1*1 convolution layer for convolution and normalize it to obtain a first convolution image, substitute the first convolution image into the 3*3 convolution layer for convolution and normalize it to obtain a second convolution image, and perform an addition operation on the first convolution image, the second convolution image and the input After the image is added, it is input into the 1*1 convolution layer and then normalized to obtain the output image. Each convolution layer in the downsampling module and the convolution module is followed by a normalization layer. The operation of the preset encoder is that the original image is downsampled 4 times. The first downsampling and the second downsampling are first passed through the downsampling module and then the convolution module. The third downsampling is first passed through the downsampling module and then through two convolution modules. The fourth downsampling is first passed through the downsampling module and then through three convolution modules. For each downsampling, the image width and height are half of the original. Sub-pixel convolution is a module in the decoder corresponding to the preset encoder, which is used to upsample the first feature map so that the final image size is consistent with the original image size. The upsampling is performed 4 times, and each upsampling will increase the image width and height to twice the original size. After each upsampling, it will pass through the convolution module.

[0075] In one implementation, the first feature map and the second feature map are concatenated to obtain the target feature map by expanding the size of the first feature map so that the size of the first feature map is the same as the size of the second feature map, and then adding the first feature map and the second feature map after the size expansion to obtain the target feature map.

[0076] In one implementation, the segmentation algorithm may be an existing segmentation algorithm such as Otsu algorithm, Canny edge detection, DeepLab2, etc.

[0077] In one embodiment, performing garbage detection on the target road image to obtain garbage data includes:

[0078] Noise filtering and contrast enhancement are performed on the target road image to obtain a first road image, and the first road image is converted into an HSV color space to obtain a second road image;

[0079] Substituting the second road image into a preset garbage detection model to obtain bounding boxes and classification results corresponding to a plurality of dry garbage;

[0080] Performing local histogram equalization on the first road image and then applying high-pass filtering to obtain a third road image;

[0081] Substitute the third road image into the preset segmentation network model to perform pixel segmentation to obtain an initial mask, and perform morphological processing on the initial mask to obtain the target wet garbage area;

[0082] The third road image is segmented according to the target wet garbage area to obtain a fourth road image, and the fourth road image is substituted into a preset deep learning classification model to obtain the type corresponding to each wet garbage area;

[0083] The bounding box and classification result corresponding to each dry garbage area and the type corresponding to each wet garbage area are recorded as garbage data.

[0084] In one implementation, by performing noise filtering and contrast enhancement on the target road image, the image quality can be improved, unnecessary interference in the image can be reduced, and the subsequent garbage detection and classification model can be helped to more accurately identify the target area and features; after conversion to the HSV color space, it helps to better separate different types of garbage in an environment with large color changes, especially for the identification of dry garbage. The HSV model is easier to maintain color consistency under specific lighting conditions than the RGB model, thereby improving the stability and reliability of detection.

[0085] In one implementation, by performing noise filtering and contrast enhancement on the target road image, the image quality can be improved, unnecessary interference in the image can be reduced, and subsequent garbage detection and classification models can be helped to more accurately identify target areas and features.

[0086] In one implementation method, the preset garbage detection model can be YOLO, Faster R-CNN, EffectiveDet, Transformer-based Models, and other commonly used garbage detection models on the market; the preset segmentation network model can be U-Net, Mask R-CNN, DeepLabv3+, etc.; the morphological operations include: Erosion: Remove possible small noise areas through corrosion operation, and remove background noise or misclassified areas; Dilation: Then use the dilation operation to expand the wet garbage area, fill the small gaps left after corrosion, and ensure the integrity of the wet garbage area; After morphological processing, a refined wet garbage area map is obtained, which can accurately represent the pixel area belonging to wet garbage in the image.

[0087] In one implementation, the preset deep learning classification model is obtained by pre-training the captured images of the wet garbage area into the Inception model.

[0088] In one implementation, local histogram equalization can enhance the local contrast of the image and make the details more prominent; high-pass filtering can remove low-frequency noise and enhance the edge features in the image. The processed image provides a basis for subsequent segmentation and recognition; by using a segmentation network model to perform pixel-level garbage area segmentation and combining morphological processing to eliminate noise, the target area of ​​wet garbage can be accurately obtained, effectively avoiding false detection and missed detection problems.

[0089] In one embodiment, determining the delivery vehicle data according to the garbage data includes:

[0090] Determine the target total amount of dry garbage based on the bounding boxes and classification results corresponding to all dry garbage, and determine the target number of dry garbage trucks to be dispatched based on the target total amount of dry garbage;

[0091] The total area of ​​the target wet garbage area is determined according to each wet garbage area, and the target wet garbage truck dispatch quantity corresponding to the total area of ​​the target wet garbage area in the historical data is obtained according to the total area of ​​the target wet garbage area.

[0092] In one implementation, by accurately calculating the quantity and category of each dry garbage area based on image recognition results, and accurately calculating the required number of dry garbage trucks to be dispatched based on these data, it can be ensured that the number of garbage trucks dispatched matches the actual needs, avoiding the situation of too many or too few garbage trucks due to estimation errors, thereby improving the efficiency of garbage collection; by accurately calculating the total area of ​​wet garbage and referring to historical data to determine the corresponding number of wet garbage trucks to be dispatched, the collection needs of wet garbage can be predicted more accurately, reducing unnecessary waste or resource shortages.

[0093] In one implementation, data on the total amount of dry garbage and the corresponding number of delivery vehicles in each previous garbage cleanup are collected, and a regression model is established based on the historical data to predict the target number of dry garbage trucks required for delivery given a total amount, wherein the target total amount of dry garbage is determined based on the bounding boxes and classification results corresponding to all dry garbage, and the volume corresponding to each dry garbage is obtained by searching a dry garbage database based on the bounding boxes and classification results corresponding to each dry garbage in the historical data; the dry garbage database stores the volume corresponding to the bounding boxes and types of each dry garbage.

[0094] In one implementation, the total area of ​​all wet garbage areas is obtained, and the wet garbage database is searched according to the total area of ​​the wet garbage areas to obtain the target wet garbage truck delivery number; the wet garbage database stores the wet garbage truck delivery numbers corresponding to different wet garbage areas.

[0095] In one implementation, accurate demand forecasting can reduce over-dispatching or under-dispatching of garbage trucks, and reduce vehicle idle rates and unnecessary fuel consumption.

[0096] In one embodiment, searching a preset database to obtain a first solution sorting set according to the garbage data and the delivery vehicle data, and determining the delivery personnel data according to the sorting of each solution in the first solution sorting set includes:

[0097] According to the target total amount of dry garbage, the target total area of ​​wet garbage area and the target number of wet garbage trucks to be delivered, a first solution sorting set is obtained by searching for a solution that satisfies the target in the preset database;

[0098] According to the total area of ​​the target wet garbage area, the target total amount of dry garbage, and the target number of dry garbage trucks to be delivered, a second solution ranking set is obtained by searching for a solution that satisfies the requirements in the preset database;

[0099] Obtain the intersection of the first solution sorting set and the second solution sorting set to obtain the target solution set. If the target solution set is not an empty set, obtain the solution with the least number of dispatched personnel in the target solution set and record it as the preset solution.

[0100] Calculate the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset plan and the target total amount of dry garbage and the target total area of ​​wet garbage. If the similarity is greater than the preset threshold, the number of people delivered in the preset plan is recorded as the delivery personnel data.

[0101] In one implementation, by searching for matching solutions based on the target total amount of dry garbage and the total area of ​​the wet garbage area, it can be ensured that the selected solution is closer to the demand in actual application, which can effectively avoid waste of resources and reduce unnecessary costs; by searching for solutions that meet the target conditions in a preset database, the most suitable solution can be quickly located, avoiding delays or errors that may occur in manual scheduling, making the scheduling of garbage trucks and personnel more accurate and efficient.

[0102] In one implementation, historical data is used to find matching solutions, which can be combined with historical experience to make more scientific and accurate decisions in a data-driven manner, eliminating subjective human judgment and reducing human errors or judgment biases. By selecting the solution with the least number of dispatched personnel, the waste of human resources can be reduced and it can be ensured that each person can play a role in the most appropriate task.

[0103] In one implementation, a preset database is searched for solutions that meet the conditions according to the target total amount of dry garbage, the target total area of ​​wet garbage areas and the target number of wet garbage trucks for delivery. Specifically, the total amount of dry garbage to be found is greater than or equal to the target total amount of dry garbage, the total area of ​​wet garbage areas to be found is greater than or equal to the target total area of ​​wet garbage areas, and the number of wet garbage trucks for delivery is equal to the target number of wet garbage trucks for delivery. The solutions that meet the conditions are then extracted from the solutions whose corresponding satisfaction levels are greater than the preset satisfaction levels and sorted according to the corresponding satisfaction levels to obtain a first sorted set of solutions. The preset satisfaction levels are determined by technical personnel. The satisfaction levels reflect how to solve the problems to be solved with the least resources. The greater the satisfaction level, the closer the resources used are to the resources for solving the problems to be solved, and vice versa.

[0104] In one implementation method, a preset database is searched for solutions that meet the conditions based on the target total area of ​​the wet garbage area, the target total amount of dry garbage and the target number of dry garbage trucks for delivery. Specifically, the total amount of dry garbage to be found is greater than or equal to the target total amount of dry garbage, the total area of ​​the wet garbage area to be found is greater than or equal to the target total area of ​​the wet garbage area, and the number of dry garbage trucks for delivery is equal to the target number of dry garbage trucks for delivery. The solutions that meet the conditions are then extracted from the solutions with corresponding satisfaction levels greater than the preset satisfaction levels, and are sorted according to the corresponding satisfaction levels to obtain a second solution sorting set.

[0105] In one implementation, if the target solution set is an empty set, the number of vehicles for delivery is relaxed by taking 1 vehicle as the measure, and the number of dry and wet vehicles is increased; the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset solution and the target total amount of dry garbage and the total area of ​​wet garbage is calculated by cosine similarity; if the similarity is less than or equal to a preset threshold, the solution with the second least number of delivery personnel in the target solution set is recorded as the preset solution, and the similarity is recalculated until the similarity is greater than the preset threshold; the preset threshold is determined by technical personnel.

[0106] In one implementation, by calculating the similarity in the target solution set, the system can select the solution that best matches the target demand, thereby ensuring the smooth progress of garbage disposal work. The similarity calculation makes the solution selection more scientific.

[0107] Based on the same inventive concept, the present invention also provides an intelligent management device for mixed crowding of people and vehicles. Figure 2 , Figure 2A schematic diagram of a structure of an intelligent management device for mixed crowding of people and vehicles provided by an embodiment of the present invention includes:

[0108] A road recognition module is used to collect environmental images in the target area and perform road recognition on the environmental images to obtain a target road image;

[0109] A garbage detection module is used to detect garbage on the target road image to obtain garbage data, and determine the delivery vehicle data according to the garbage data;

[0110] A personnel dispatch determination module is used to search a preset database to obtain a first scheme sorting set according to the garbage data and the dispatch vehicle data, and determine the dispatch personnel data according to the sorting of each scheme in the first scheme sorting set;

[0111] The delivery plan determination module is used to determine the delivery plan based on the delivery vehicle data and delivery personnel data.

[0112] Based on an intelligent management device for mixed traffic of people and vehicles provided by an embodiment of the present invention, by performing garbage detection on the target road image and generating garbage data, the degree of dirtiness and type of garbage in the area can be accurately assessed. The system can determine the number and type of sanitation vehicles that need to be dispatched, and further determine the number and tasks of the dispatch personnel based on the dispatched vehicle data, thereby ensuring the reasonable allocation of resources, thereby optimizing the allocation of resources, avoiding the problem of over-delivery or under-delivery, and improving cleaning efficiency.

[0113] In one embodiment, the road recognition module includes:

[0114] A road recognition module is used to preprocess the environment image to obtain an initial image, perform grayscale conversion on the initial image to obtain a target grayscale image, and substitute the target grayscale image into a preset encoder to perform feature extraction to obtain a first feature map;

[0115] An image stitching module, used for processing the first feature map by sub-pixel convolution to obtain a second feature map, and stitching the first feature map and the second feature map to obtain a target feature map;

[0116] The road recognition module is used to process the target feature map through a segmentation algorithm to obtain a road distribution area, and cut the environment image according to the road distribution area to obtain an image containing the road distribution area and record it as a target road image.

[0117] In one embodiment, the spam detection module includes:

[0118] An image enhancement module, used for performing noise filtering and contrast enhancement on the target road image to obtain a first road image, and converting the first road image into an HSV color space to obtain a second road image;

[0119] A dry garbage detection module, used to substitute the second road image into a preset garbage detection model to obtain bounding boxes and classification results corresponding to a plurality of dry garbage;

[0120] An image filtering module, used for performing local histogram equalization on the first road image and then applying high-pass filtering to obtain a third road image;

[0121] A wet garbage area determination module is used to substitute the third road image into a preset segmentation network model to perform pixel segmentation to obtain an initial mask, and perform morphological processing on the initial mask to obtain a target wet garbage area;

[0122] A wet garbage detection module, used to segment the third road image according to the target wet garbage area to obtain a fourth road image, and substitute the fourth road image into a preset deep learning classification model to obtain a type corresponding to each wet garbage area;

[0123] The garbage data determination module is used to record the border and classification result corresponding to each dry garbage and the type corresponding to each wet garbage area as garbage data.

[0124] In one embodiment, the spam detection module further includes:

[0125] A dry garbage truck quantity determination module is used to determine the target total amount of dry garbage based on the bounding boxes and classification results corresponding to all dry garbage, and determine the target number of dry garbage trucks to be dispatched based on the target total amount of dry garbage;

[0126] The wet garbage truck quantity determination module is used to determine the total area of ​​the target wet garbage area according to each wet garbage area, and obtain the target wet garbage truck dispatch quantity corresponding to the total area of ​​the target wet garbage area in the historical data according to the total area of ​​the target wet garbage area.

[0127] In one embodiment, the personnel dispatch determination module includes:

[0128] A first scheme ranking set determination module is used to find a scheme that satisfies the target total amount of dry garbage, the target total area of ​​wet garbage area and the target number of wet garbage trucks to be delivered to obtain a first scheme ranking set;

[0129] A second scheme ranking set determination module is used to find a scheme that satisfies the requirements in a preset database according to the total area of ​​the target wet garbage area, the target total amount of dry garbage, and the target number of dry garbage trucks to be delivered, to obtain a second scheme ranking set;

[0130] A preset solution determination module is used to obtain the intersection of the first solution sorting set and the second solution sorting set to obtain a target solution set. If the target solution set is not an empty set, the solution with the least number of dispatched personnel in the target solution set is obtained and recorded as the preset solution.

[0131] The similarity calculation module is used to calculate the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset plan and the target total amount of dry garbage and the total area of ​​wet garbage. If the similarity is greater than the preset threshold, the number of people delivered in the preset plan is recorded as the delivery personnel data.

[0132] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A smart management method for mixed crowding of people and vehicles, characterized in that: The method comprises: Collecting an environmental image in a target area, and performing road recognition on the environmental image to obtain a target road image; Performing garbage detection on the target road image to obtain garbage data, and determining dispatch vehicle data according to the garbage data; Searching a preset database according to the garbage data and the delivery vehicle data to obtain a first solution sorting set, and determining the delivery personnel data according to the sorting of each solution in the first solution sorting set; A delivery plan is determined based on the delivery vehicle data and the delivery personnel data.

2. The intelligent management method for mixed crowding of people and vehicles according to claim 1 is characterized in that: Performing road recognition on the environment image to obtain a target road image includes: Preprocessing the environment image to obtain an initial image, performing grayscale conversion on the initial image to obtain a target grayscale image, substituting the target grayscale image into a preset encoder to perform feature extraction to obtain a first feature map; Processing the first feature map by sub-pixel convolution to obtain a second feature map, and concatenating the first feature map and the second feature map to obtain a target feature map; The target feature map is processed by a segmentation algorithm to obtain a road distribution area, and the environment image is cut according to the road distribution area to obtain an image containing the road distribution area, which is recorded as a target road image.

3. The intelligent management method for mixed crowding of people and vehicles according to claim 1 is characterized in that: Performing garbage detection on the target road image to obtain garbage data includes: Performing noise filtering and contrast enhancement on the target road image to obtain a first road image, and converting the first road image into an HSV color space to obtain a second road image; Substituting the second road image into a preset garbage detection model to obtain bounding boxes and classification results corresponding to a plurality of dry garbage; Performing local histogram equalization on the first road image and then applying high-pass filtering to obtain a third road image; Substituting the third road image into a preset segmentation network model to perform pixel segmentation to obtain an initial mask, and performing morphological processing on the initial mask to obtain a target wet garbage area; Segmenting the third road image according to the target wet garbage area to obtain a fourth road image, and substituting the fourth road image into a preset deep learning classification model to obtain a type corresponding to each wet garbage area; The bounding box and classification result corresponding to each dry garbage area and the type corresponding to each wet garbage area are recorded as garbage data.

4. The intelligent management method for mixed crowding of people and vehicles according to claim 3 is characterized in that: Determining the delivery vehicle data according to the garbage data includes: Determine the target total amount of dry garbage based on the bounding boxes and classification results corresponding to all dry garbage, and determine the target number of dry garbage trucks to be dispatched based on the target total amount of dry garbage; The total area of ​​the target wet garbage area is determined according to each wet garbage area, and the target wet garbage truck dispatch quantity corresponding to the total area of ​​the target wet garbage area in the historical data is obtained according to the total area of ​​the target wet garbage area.

5. The intelligent management method for mixed crowding of people and vehicles according to claim 4 is characterized in that: Searching a preset database according to the garbage data and the delivery vehicle data to obtain a first solution sorting set, and determining the delivery personnel data according to the sorting of each solution in the first solution sorting set includes: According to the target total amount of dry garbage, the target total area of ​​wet garbage area and the target number of wet garbage trucks to be delivered, a first solution sorting set is obtained by searching for a solution that satisfies the target in the preset database; According to the total area of ​​the target wet garbage area, the target total amount of dry garbage, and the target number of dry garbage trucks to be delivered, a second solution ranking set is obtained by searching for a solution that satisfies the requirements in the preset database; Obtain the intersection of the first solution sorting set and the second solution sorting set to obtain a target solution set. If the target solution set is not an empty set, obtain the solution with the least number of dispatched personnel in the target solution set and record it as a preset solution. Calculate the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset plan and the target total amount of dry garbage and the target total area of ​​wet garbage. If the similarity is greater than a preset threshold, the number of people delivered in the preset plan is recorded as delivery personnel data.

6. An intelligent management device for mixed traffic of people and vehicles, characterized in that: The device comprises: A road recognition module is used to collect an environmental image in a target area and perform road recognition on the environmental image to obtain a target road image; A garbage detection module, used for performing garbage detection on the target road image to obtain garbage data, and determining the dispatch vehicle data according to the garbage data; A personnel dispatch determination module, configured to search a preset database according to the garbage data and the dispatch vehicle data to obtain a first scheme sorting set, and determine the dispatch personnel data according to the sorting of each scheme in the first scheme sorting set; A delivery plan determination module is used to determine the delivery plan based on the delivery vehicle data and the delivery personnel data.

7. The intelligent management device for mixed traffic of people and vehicles according to claim 6, characterized in that: The road recognition module comprises: A road recognition module, used for preprocessing the environment image to obtain an initial image, performing grayscale conversion on the initial image to obtain a target grayscale image, and substituting the target grayscale image into a preset encoder to perform feature extraction to obtain a first feature map; An image stitching module, configured to process the first feature map by sub-pixel convolution to obtain a second feature map, and stitch the first feature map and the second feature map to obtain a target feature map; The road recognition module is used to process the target feature map through a segmentation algorithm to obtain a road distribution area, and cut the environment image according to the road distribution area to obtain an image containing the road distribution area and record it as a target road image.

8. The intelligent management device for mixed traffic of people and vehicles according to claim 6, characterized in that: The garbage detection module includes: An image enhancement module, configured to perform noise filtering and contrast enhancement on the target road image to obtain a first road image, and convert the first road image into an HSV color space to obtain a second road image; A dry garbage detection module, used to substitute the second road image into a preset garbage detection model to obtain a plurality of bounding boxes and classification results corresponding to dry garbage; An image filtering module, configured to perform local histogram equalization on the first road image and then apply high-pass filtering to obtain a third road image; A wet garbage area determination module is used to substitute the third road image into a preset segmentation network model to perform pixel segmentation to obtain an initial mask, and perform morphological processing on the initial mask to obtain a target wet garbage area; A wet garbage detection module, configured to segment the third road image according to the target wet garbage area to obtain a fourth road image, and substitute the fourth road image into a preset deep learning classification model to obtain a type corresponding to each wet garbage area; The garbage data determination module is used to record the border and classification result corresponding to each dry garbage and the type corresponding to each wet garbage area as garbage data.

9. The intelligent management device for mixed traffic of people and vehicles according to claim 8, characterized in that: The garbage detection module also includes: A dry garbage truck quantity determination module is used to determine the target total amount of dry garbage according to the bounding boxes and classification results corresponding to all dry garbage, and determine the target dry garbage truck dispatch quantity according to the target total amount of dry garbage; The wet garbage truck quantity determination module is used to determine the total area of ​​the target wet garbage area according to each wet garbage area, and obtain the target wet garbage truck dispatch quantity corresponding to the total area of ​​the target wet garbage area in the historical data according to the total area of ​​the target wet garbage area.

10. The intelligent management device for mixed traffic of people and vehicles according to claim 9, characterized in that: The personnel dispatch determination module includes: A first scheme ranking set determination module is used to find a scheme that satisfies the target total amount of dry garbage, the target total area of ​​wet garbage area and the target number of wet garbage trucks to be delivered to obtain a first scheme ranking set; A second scheme ranking set determination module is used to find a scheme that satisfies the requirements in a preset database according to the total area of ​​the target wet garbage area, the target total amount of dry garbage, and the target number of dry garbage trucks to be delivered, to obtain a second scheme ranking set; A preset solution determination module is used to obtain the intersection of the first solution sorting set and the second solution sorting set to obtain a target solution set. If the target solution set is not an empty set, the solution with the least number of dispatched personnel in the target solution set is obtained and recorded as the preset solution. The similarity calculation module is used to calculate the similarity between the total amount of dry garbage and the total area of ​​wet garbage in the preset plan and the target total amount of dry garbage and the total area of ​​wet garbage. If the similarity is greater than a preset threshold, the number of people delivered in the preset plan is recorded as delivery personnel data.

Images