Intelligent method for identifying types and potential of outdoor fitness spaces along streets

Abstract

The application discloses a method for intelligently identifying the type and potential of outdoor fitness space beside a street, relates to the technical field of city planning, landscape gardening and artificial intelligence, and comprises the following steps: inputting the city space image data set of a target region into a pre-established outdoor fitness space activity type identification model to obtain an outdoor fitness space potential point corresponding to the outdoor fitness space activity type of the city space image; and inputting the city space image data set of the target region into a corresponding outdoor fitness space activity potential evaluation model to obtain the activity potential evaluation of the outdoor fitness space potential point corresponding to the outdoor fitness space activity type of the city space image; the application can analyze the city space site by means of intelligent target detection and deep learning technology, quantize the configuration relationship between the fitness behavior of the space site and the characteristics of the site elements, and establish a stable derivation relationship between the complete visual scene of the city space and the outdoor fitness activity potential.

Description

Technical Field

[0001] This invention relates to the fields of urban planning, landscape architecture and artificial intelligence technology, specifically a method for intelligently identifying the type and potential of outdoor fitness spaces along streets. Background Technology

[0002] Currently, the industry primarily relies on manual analysis for evaluating urban fitness spaces. This evaluation often employs literature review, using successful case studies of fitness spaces in developed countries to analyze their practices and summarize their experiences; or it uses logical analysis, applying relevant theories to analyze the site. While artificial intelligence and big data have been incorporated into the analysis of urban fitness spaces within the urban public space planning industry, the research objects are mostly existing fitness spaces, analyzing their distribution characteristics. For example, a region might use QGIS spatial research and analysis tools, with POI information provided by Gaode Maps as the research object, to analyze the distribution characteristics of outdoor recreational fitness spaces. This demonstrates that while there have been some attempts at data-driven analysis within the industry, this data analysis remains superficial.

[0003] In systems relying on manual evaluation, inefficiency and inaccurate results are common problems. Currently, the theoretical framework for establishing fitness spaces remains incomplete. Due to their inherent randomness and numerous influencing factors, summarizing relevant patterns is challenging. Manual analysis also depends on the planner's capabilities, and differing perspectives among planners lead to significant uncertainty in the results. Existing data-driven methods also have many shortcomings. Analysis of urban spatial data often remains focused on a single aspect, resulting in superficial analysis lacking comprehensiveness. Current data-driven analyses of fitness spaces primarily focus on existing spaces, lacking foresight. Summary of the Invention

[0004] To address the shortcomings mentioned in the background art, the present invention aims to provide a method for intelligently identifying the type and potential of outdoor fitness spaces along streets. This method can analyze urban spaces using intelligent target detection and deep learning technologies, quantify the configuration relationship between fitness behaviors and site element characteristics, and establish a stable deductive relationship between the complete visual scene of urban space and the potential of outdoor fitness activities.

[0005] The objective of this invention can be achieved through the following technical solution: a method for intelligently identifying the type and potential of outdoor fitness spaces along streets, comprising: acquiring a dataset of urban spatial images of the target area;

[0006] The urban spatial image dataset of the target area is input into the pre-established outdoor fitness space activity type identification model. Through the feature element identification system of outdoor fitness space activity type identification, the potential outdoor fitness space points corresponding to the outdoor fitness space activity type in the urban spatial image are obtained.

[0007] The urban spatial image dataset of the target area is input into a pre-established outdoor fitness space activity potential evaluation model corresponding to the outdoor fitness space activity type, so as to obtain the activity potential evaluation of the outdoor fitness space potential points corresponding to the outdoor fitness space activity type in the urban spatial image.

[0008] Based on the activity potential evaluation of outdoor fitness space potential points corresponding to the activity types of outdoor fitness spaces, target areas are selected for outdoor fitness venues corresponding to the activity types of outdoor fitness spaces.

[0009] Furthermore, an urban spatial image dataset of the target area is obtained, including:

[0010] The process involves acquiring the driving trajectory map of the street view data collection vehicle from the map open platform within the target area, vectorizing the trajectory map in the geographic information system platform, determining the collection interval according to requirements, converting the vector line data into point data, obtaining urban spatial street view collection points, acquiring the surrounding urban spatial images corresponding to all urban spatial street view collection points, and constructing an urban spatial image dataset; the surrounding urban spatial images include complete visual image information and geospatial location information.

[0011] Furthermore, the pre-established outdoor fitness space activity type identification model includes:

[0012] Obtain images containing feature elements from the existing feature element set within the outdoor fitness space activity type identification system. Then, select and label the feature elements within the outdoor fitness space activity type identification system from the images containing feature elements from the existing feature element set within the outdoor fitness space activity type identification system, and use this as a training set for outdoor fitness space activity type identification.

[0013] Using a ResNet50 network built on a deep learning framework as the underlying network and a Faster RCNN network for object detection, an outdoor fitness space type identification model was constructed. Training samples from the outdoor fitness space activity type identification training set were used as input for training. Feature elements within the outdoor fitness space activity type identification system were identified in the training samples until the accuracy of the outdoor fitness space activity type identification model met the requirements, thus completing the establishment of the outdoor fitness space activity type identification model.

[0014] Furthermore, the construction of an outdoor fitness space activity type identification system includes:

[0015] The image dataset of outdoor fitness activity spaces where actual fitness activities exist extracts the feature elements of the fitness activity space corresponding to each type of outdoor fitness activity space, and obtains the feature element set of the fitness activity space corresponding to each type of outdoor fitness activity space.

[0016] The characteristic elements of fitness activity spaces are classified into decisive positive decision-making characteristic elements, general positive decision-making characteristic elements, and decisive negative decision-making characteristic elements according to their degree of conformity to the corresponding outdoor fitness space activity type.

[0017] By identifying the feature elements in the outdoor fitness space activity type identification system, the potential points of outdoor fitness space corresponding to the type of outdoor fitness space activity in the urban space image are obtained. This includes: if the pre-established outdoor fitness space activity type identification model identifies the decisive positive decision elements corresponding to the outdoor fitness activity or only the general positive decision elements, then this point is taken as the potential point of outdoor fitness space corresponding to the type of outdoor fitness space activity in the urban space image.

[0018] If no decisive positive decision-making element for the corresponding outdoor fitness activity is identified, but no decisive negative decision-making element for the corresponding outdoor fitness activity is identified, or if no decisive positive decision-making element, decisive negative decision-making element, or general positive decision-making element for the corresponding outdoor fitness activity is identified, then this location will not be considered as a potential outdoor fitness space point corresponding to the type of outdoor fitness space activity in the urban spatial image.

[0019] Furthermore, the positive decision-making characteristic elements, general positive decision-making characteristic elements, and decisive negative decision-making characteristic elements include one or more of the following factors: site size, site shape, site paving, site greenery, fitness facilities, and municipal facilities.

[0020] Furthermore, the outdoor fitness activity space image dataset that actually exists includes: an outdoor fitness activity space image dataset for each corresponding type of outdoor fitness activity and a set of outdoor fitness actual space image datasets.

[0021] Furthermore, the acquisition of the outdoor fitness space image dataset for each corresponding outdoor fitness space activity type includes:

[0022] By using the map open platform to collect and record geospatial location information, and by calling the static panoramic image service of the map open platform's application programming interface, we can obtain spatial images of actual outdoor fitness activities and establish a dataset of spatial images of actual outdoor fitness activities corresponding to different types of outdoor fitness activities.

[0023] Furthermore, the acquisition of the actual outdoor fitness space image dataset for each corresponding type of outdoor fitness space activity includes: a set of actual outdoor space street view images and a set of actual outdoor space cropped images;

[0024] The acquisition of the actual spatial street view image set includes: using the actual outdoor fitness spaces that are spontaneously shared by users on multi-source social media platforms, where activities are taking place and their spatial geographic location information can be identified, and by calling the static panoramic image service of the map open platform's application programming interface, to obtain the actual outdoor fitness space street view images and establish the actual spatial street view image set.

[0025] The acquisition of the actual spatial image set includes: for outdoor fitness spaces where the geospatial location information cannot be determined, the method of directly capturing platform images is selected to obtain the actual spatial image set of outdoor fitness spaces.

[0026] Furthermore, the pre-established outdoor fitness space activity potential evaluation model corresponding to the type of outdoor fitness space activity includes:

[0027] Using the outdoor fitness space activity potential evaluation system as the evaluation standard, outdoor fitness space images corresponding to the type of outdoor fitness space activity are selected from the outdoor fitness space image dataset of actual outdoor fitness activities as training samples for activity potential labeling, serving as the training set for evaluating the activity potential of each type of outdoor fitness activity.

[0028] Based on deep learning frameworks, a fully connected image convolutional neural network was built, taking as input a training set for evaluating the potential of each type of outdoor fitness activity and a corresponding outdoor fitness space activity.

[0029] Deep learning is used to train the activity types until they meet the accuracy requirements, thereby obtaining an evaluation model for the potential of outdoor fitness space activities of each type.

[0030] Furthermore, the construction of the outdoor fitness space activity potential evaluation system includes:

[0031] Annotate the activity potential of outdoor fitness activity spatial images for each corresponding type of outdoor fitness activity in the dataset of outdoor fitness activity spatial images where actual fitness activities actually exist.

[0032] Activity potential indicators include:

[0033] The number of descriptive points in each outdoor fitness space activity potential evaluation system was summarized, and the corresponding outdoor fitness activity space images were compared one by one according to the descriptive points. The scores were calculated by cumulative summation. If a descriptive point was met, one point was awarded; if not, no points were awarded. Based on this principle, all training samples were scored as potential labels for each corresponding outdoor fitness activity space image.

[0034] Urban spatial images are input into a pre-established outdoor fitness space activity potential evaluation model corresponding to the type of outdoor fitness space activity. This yields an outdoor fitness space activity potential score for that type of activity. Under the same type of outdoor fitness space activity, K-means clustering is performed on the outdoor fitness space activity potential scores of the urban spatial images of the target area to obtain an outdoor fitness space activity potential level. This level is used to describe the evaluation result of the potential of that type of outdoor fitness space activity.

[0035] Beneficial effects:

[0036] It can use intelligent target detection and deep learning technology to analyze urban spaces, quantify the configuration relationship between fitness behavior and site characteristics, and establish a stable inference relationship between the complete visual scene of urban space and the potential of outdoor fitness activities; thereby improving the efficiency and accuracy of urban fitness space exploration. Attached Figure Description

[0037] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort:

[0038] Figure 1 This is a flowchart of the method of the present invention;

[0039] Figure 2 This is a schematic diagram of a method according to one embodiment of the present invention;

[0040] Figure 3 This is a flowchart illustrating one embodiment of the present invention;

[0041] Figure 4 This is an image of an outdoor fitness activity space related to basketball, as described in this invention.

[0042] Figure 5 This is another outdoor fitness activity space image related to basketball in this invention. Detailed Implementation

[0043] The technical solutions in this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described examples are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] like Figure 1 and Figure 2As shown, the method for intelligently identifying the type and potential of outdoor fitness spaces along the street includes the following steps:

[0045] Construct an urban spatial image dataset of the target area;

[0046] The urban spatial image dataset for the target area consists of urban street view images of the target city collected based on a map open platform. The specific steps are as follows:

[0047] The process involves acquiring the driving trajectory of a street view data collection vehicle within the target area from a map open platform. This trajectory is then vectorized within a geographic information system platform, and the data collection interval is determined based on research needs. The vector line data is converted into point data to obtain urban spatial street view data collection points. Next, the static panoramic image service of the map open platform's application programming interface (API) is called in Python. Based on the urban spatial street view data collection points and adjusted according to research needs, parameters are modified to obtain urban spatial images, thus constructing an urban spatial image dataset. This dataset contains complete visual images and geospatial location information for each urban street view collection point.

[0048] The reasons for choosing street view images from the Map Open Platform to construct the urban spatial image dataset for the target area are as follows: 1) The panoramic image service of the Map Open Platform is currently the most widely known urban environmental image data source with geospatial information that can be acquired in batches, which can meet the needs of this technology for large-scale urban applications; 2) The purpose of this technology is to intelligently identify the type and potential evaluation of outdoor fitness spaces along the street. The panoramic image service of the Map Open Platform can personalize the acquisition parameters of the street view acquisition vehicle to meet the multi-angle image acquisition needs of urban streets, which matches the goal of this technology.

[0049] In this example, the main urban area is used as the target area, and Baidu Maps is used as the selected map open platform. First, electronic map capture software is used to capture the urban navigation trajectory map of the planned area in Baidu Maps.

[0050] The trajectory map was vectorized using the geographic information system platform ArcGIS. After converting the vector lines to points, urban spatial street view collection points were obtained. The sampling distance was set to 50m, which ensures that the street view images of two adjacent collection points can completely describe the street spatial information without omission.

[0051] Determine the urban spatial image acquisition parameters, crawl the near-street street view images at three angles (45°, 90°, and 135°) of the street view vehicle's driving direction for each urban spatial street view acquisition point, and record the corresponding geospatial coordinates;

[0052] Urban spatial image dataset The data is input into a pre-established outdoor fitness space activity type identification model, and the potential outdoor fitness space points corresponding to the urban space image are obtained through feature element identification and combination.

[0053] The process of constructing a complete model for identifying activity types in outdoor fitness spaces includes:

[0054] A dataset of outdoor fitness activity space images that actually exist is constructed, and an outdoor fitness activity type identification system is constructed. The outdoor fitness activity type identification system is obtained by summarizing and extracting outdoor fitness activity space image data.

[0055] An outdoor fitness space activity type identification training set was constructed. The ResNet50 network, built based on a deep learning framework tool, was used as the underlying network, and the Faster RCNN network was used as the object detection method to build the outdoor fitness space activity type identification model. The training was conducted using training samples from the outdoor fitness space activity type identification training set. The training samples in the outdoor fitness space activity type identification training set were collected and labeled based on the feature elements contained in the outdoor fitness activity type identification system.

[0056] The process of constructing an image set of outdoor fitness activity spaces that actually exist is as follows:

[0057] Suppose there are N types of outdoor fitness activities (N=1,2,3,...,N, N∈N) + Establish an image set of outdoor fitness activity spaces where actual fitness activities exist. Define the x-th (1 <= x <= N, i ∈ N) + The image set of outdoor fitness activity spaces for households is ,exist:

[0058]

[0059] Where the x-th (1<=x<=N, i∈N) + The image set of outdoor fitness activity spaces for households is Includes a dataset of images of actual outdoor fitness activity spaces. Image datasets of actual outdoor fitness spaces ;

[0060] Among them, the outdoor fitness actual activity space image dataset The actual outdoor fitness activity space in the context of outdoor fitness activity x can be understood as follows: For example, for outdoor fitness activity x, the actual outdoor fitness activity space image of outdoor fitness activity x is obtained by using the map open platform to pick and record geospatial location information based on the known spatial location of the outdoor fitness activity, and by calling the static panoramic image service of the map open platform's application programming interface (API) through Python to obtain the actual outdoor fitness activity space image, thus establishing the outdoor fitness activity space image dataset for outdoor fitness activity x. ;

[0061] Outdoor fitness actually exists in spatial image datasets The actual outdoor fitness space in the context of outdoor fitness activities can be understood as follows: for example, for outdoor fitness activity x, the actual outdoor fitness space image of outdoor fitness activity x is obtained by utilizing user-shared images from multiple social media platforms, showing ongoing activities and verifiable spatial geographic locations. This is achieved by using Python to call the static panoramic image service of the map open platform's application programming interface (API) to obtain street view images of the actual outdoor fitness space, thus establishing a street view image set of the actual outdoor fitness space. Among them, the actual space for outdoor fitness activities that are spontaneously shared by users on multiple social media platforms, are ongoing, and whose spatial geospatial location information can be identified, includes videos and images of exercise activities posted by users on Douyin (TikTok) that carry geospatial location information or textual information specifically describing geospatial location, and videos and images of exercise activities posted by users on Weibo (Weibo) that carry geospatial location information or textual information specifically describing geospatial location.

[0062] For outdoor fitness spaces whose geospatial location information could not be determined, images of the actual outdoor fitness spaces were obtained by directly capturing images from the platform, and an image set of these actual spaces was established. The set of street view images of actual outdoor fitness spaces and the set of cropped images of actual outdoor fitness spaces together constitute the set of images of actual outdoor fitness spaces, and the relationship is as follows:

[0063]

[0064] Then we have:

[0065]

[0066] The process of constructing an outdoor fitness space activity type identification system is as follows:

[0067] Based on outdoor fitness activity space image dataset The characteristic elements of each outdoor fitness activity space are extracted and logically summarized to obtain a set of characteristic elements for each fitness activity space. The set includes decisive positive decision-making characteristic elements, general positive decision-making characteristic elements, and decisive negative decision-making characteristic elements.

[0068] All three elements are established based on the "Selection Standards for National Fitness Venues and Facilities". The decisive positive decision-making characteristic element is one that has a significant positive impact on the occurrence of the fitness activity; that is, as long as the element exists, the corresponding activity can occur, such as a basketball hoop for basketball and fitness facilities for individual fitness activities. The general positive decision-making element is one that has a positive impact on the occurrence of the fitness activity, but its positive impact on the specific sport is less than that of the decisive positive element. It refers to the basic conditions that can meet the needs of the sport and have the potential to stimulate activity, such as a regular and flat surface, which may also lead to basketball and individual fitness activities. The decisive negative decision-making characteristic element is one that has a significant negative impact on the occurrence of the fitness activity; that is, in the absence of a basketball hoop, the activity will not occur. In targeted activity facilities and venues with average basic conditions, there are elements that seriously hinder sports activities, such as sand for basketball and sand for skateboarding. These three factors are used to determine the potential of outdoor fitness spaces. Since decisive positive decision-making features have a strong orientation towards the type of sport, and the identification of these features in the technical solution is based on real urban spaces, their existence inevitably reflects the actual needs of users. Therefore, they can compensate for or weaken the negative impact of decisive negative decision-making features on sports. General positive decision-making features, on the other hand, can only meet the basic venue requirements for sports and are more easily interfered with by decisive negative decision-making features. Therefore, the final priority of the three is: decisive positive decision-making features > decisive negative decision-making features > general positive decision-making features.

[0069] Positive decision-making characteristic elements, general positive decision-making characteristic elements, and decisive negative decision-making characteristic elements include one or more of the following factors: site size, site shape, site paving, site greenery, fitness facilities, and municipal facilities.

[0070] Let the feature set of the x-th type of fitness activity space be... The set of decisive positive decision-making characteristic elements is The general set of positive decision-making feature elements is as follows: The set of decisive negative decision-making characteristic elements is Then we have:

[0071]

[0072]

[0073]

[0074]

[0075] Wherein, px represents the decisive positive decision-making feature element of outdoor fitness activity x, and a is its number; gx represents the general positive decision-making feature element of outdoor fitness activity x, and b is its number; ex represents the decisive negative feature element of outdoor fitness activity x, and c is its number.

[0076] In this example, The motions included are as follows:

[0077] 1.type_football; 2.type_plaza dancing; 3.type_children's activities; 4.type_individual fitness activities; 5.type_basketball; 6.type_volleyball; 7.type_badminton; 8.type_skate boarding; 9.type_roller skating; 10.type_linear-fitness-activities.

[0078] Taking individual fitness activities (4.type_individual fitness activities), basketball (5.type_basketball), and skateboarding (8.type_skate boarding) as examples, the type identification framework is as follows:

[0079] For type 4.individual fitness activities, the decisive positive decision factor is fitness facilities; the decisive negative decision factor is road lines; and the general positive decision factors are brick ground, rubber ground, cement ground, and grass.

[0080] In this example, x=4, a=1, b=1, c=4;

[0081] ;

[0082] Where p41 = fitness facilities; e41 = road lines; g41 = brick ground; g42 = plastic ground; g43 = cement ground; g44 = grass.

[0083] For 5.type_basketball, the decisive positive decision element is the basketball hoop; the decisive negative decision elements are sand, grass, water, steps, zebra crossing, and road lines; the general positive decision elements are block concrete field, block plastic field, lighting facilities, and seating facilities.

[0084] In this example, x=5, a=1, b=6, c=4;

[0085] ;

[0086] Where p51 = basketball hoop; e51 = sand, e52 = grass, e53 = water, e54 = steps, e55 = zebra crossing, e56 = road line; g51 = block cement field, g52 = block plastic field, g53 = lighting facilities, g54 = seating facilities.

[0087] For 8.type_skate boarding, there are no decisive positive decision factors; the decisive negative decision factors are sand, grassland, road lines, and vehicle clusters; the general positive decision factors are blocky cement ground and plastic ground.

[0088] In this example, x=8, a=0, b=4, c=2;

[0089] ;

[0090] Among them, e81 = sandy land, e82 = grassland, e83 = road line, e84 = vehicle cluster; g81 = cement ground, g82 = plastic ground.

[0091] The process of constructing a training set for identifying activity types in outdoor fitness spaces is as follows:

[0092] Construct a feature element image dataset and label the image samples with feature elements; the feature element image dataset is collected and classified based on the feature elements contained in the outdoor fitness space activity type identification system, and the data source is open image data on the Internet. The number of samples in each feature element image dataset must be consistent.

[0093] Define the set of integrated spatial feature elements as E, ;

[0094] By merging duplicate elements in the comprehensive spatial feature set and recoding the elements in E, we have: , where k is the total number of spatial feature elements;

[0095] By collecting images from the internet, we obtain images containing the j-th ( Image of 1 spatial feature element Construct its feature image dataset Assuming the set contains m images, then:

[0096]

[0097] The feature annotation process uses the LabelImg tool to manually select and annotate the feature elements in the feature element image dataset to obtain labeled samples. The labeled samples are then compiled as a training set for identifying the activity type of outdoor fitness space.

[0098] The process of building an outdoor fitness space type identification model using a ResNet50 network built on a deep learning framework as the underlying network and a Faster R-CNN network as the object detection method is as follows:

[0099] We selected Faster RCNN as the object detection framework, constructed an outdoor fitness type identification model, preprocessed all input spatial feature element images, used ResNet50 based on residual network as the feature extraction network, established RPN network to generate region selection boxes, and established Classifier classification network to classify the images within the candidate boxes according to features.

[0100] In terms of model optimization, the Adam optimizer is used, which adds an adaptive learning rate adjustment step to the stochastic gradient descent method to obtain better training results.

[0101] Finally, the model training results are evaluated using parameters such as average precision (mAP), precision, and recall.

[0102] A portion of the data from the outdoor fitness space activity type identification training set was randomly selected for training, while the remainder was used for model testing. This ensured the randomness of the model training and the universality and generalizability of the model's performance.

[0103] After establishing a complete target discrimination training set, a Python program was used to randomly generate a partitioning method. 90% of the samples from the complete outdoor fitness space dataset were randomly selected to form the training set, and the remaining 10% formed the test set. In the training set, 90% of the randomly selected samples were used for parameter tuning during training, and the remaining 10% were used to validate the accuracy during training, allowing for real-time evaluation of the model's generalization ability. Finally, a k-feature element discrimination model was constructed. , , , ..., The k feature element identification models together constitute the outdoor fitness space activity type identification model.

[0104] The potential outdoor fitness spaces corresponding to urban spatial images are obtained by identifying and combining feature elements. The process is as follows:

[0105] Urban spatial image dataset Input to each feature element discrimination model middle( ), perform feature element detection, and if the decisive positive decision-making factors of outdoor fitness activity x are identified. Or only identify general positive decision-making factors. If the site meets the spatial requirements of outdoor fitness activity x, then the geographic space corresponding to the urban spatial image is determined to be the corresponding outdoor fitness site x.

[0106] If the decisive positive decision-making factors for outdoor fitness activity x are not identified Furthermore, it identified the decisive negative decision-making factors. or no relevant elements ( If the site is detected, it does not meet the spatial requirements for outdoor fitness activity x, meaning that the geographic space corresponding to the urban spatial image is not determined to be an outdoor fitness site x.

[0107] In this example, taking the spatial potential point identification of 5.type_basketball as an example, if the input urban spatial image, after applying the outdoor fitness activity type identification model, yields the following result, then the space is an outdoor fitness venue for basketball activities:

[0108] 1) Identify p51;

[0109] 2) Only one or more of g51, g52, g53, and g54 can be identified;

[0110] If the result is as follows, then the space is not an outdoor fitness area for basketball activities:

[0111] 1) p51 was not identified, but one or more of e51, e52, e53, e54, e55, and e56 were identified;

[0112] 2) None of p51, e51, e52, e53, e54, e55, e56, g51, g52, g53, g54, g55, or g56 were identified;

[0113] like Figure 2 As shown, the method for intelligently identifying the type and potential of outdoor fitness spaces along the street includes the following steps in its outdoor fitness space potential evaluation method:

[0114] The complete process of constructing a pre-established deep learning-based outdoor fitness space activity potential evaluation model includes: building an outdoor fitness space activity potential evaluation system; establishing an outdoor fitness space activity potential evaluation training set; and training the deep learning-based outdoor fitness space activity potential evaluation model.

[0115] Based on the "Selection Standards for National Fitness Venues and Facilities", an evaluation system for the potential of outdoor fitness spaces is constructed. This system is a collection of potential evaluation standards for all types of outdoor fitness spaces. Each type of potential evaluation standard consists of a series of potential description points. These description points describe the site characteristics that an excellent outdoor fitness activity site should meet, including one or more aspects of site space conditions, site landscape conditions, site facility conditions, and site activity conditions.

[0116] In this example, taking basketball (5.type_basketball) as an example, the key descriptive points included in its potential evaluation system are:

[0117] 1. An open space without any obstructions;

[0118] 2. The activity area should be paved with a smooth, even surface;

[0119] 3. Cement or plastic surfaces;

[0120] 4. An activity area enclosed on three sides by buildings or trees;

[0121] 5. It has public service facilities such as lighting fixtures and rest seats;

[0122] 6. It has recreational facilities such as basketball hoops;

[0123] 7. A safe area with no vehicles parked;

[0124] 8. Sites that are not easily disturbed and have no road intersections or building entrances.

[0125] The complete process of establishing a training set for evaluating the activity potential of outdoor fitness spaces includes:

[0126] Training image samples were collected from various types of outdoor fitness activity spaces. The image samples were derived from an outdoor fitness activity space image dataset. ,definition For the training set of outdoor fitness activity x, we have:

[0127]

[0128] Then we have:

[0129] ;

[0130] Using the outdoor fitness space activity potential evaluation system as the evaluation standard, each type of outdoor fitness potential evaluation image set was evaluated. The training samples were labeled with activity potential. The labeling process was as follows: the number of descriptive points in the evaluation system for activity potential of each type of outdoor fitness space was summarized, and the corresponding outdoor fitness activity space images were compared one by one according to the descriptive points. The scores were calculated by cumulative summation. If a descriptive point was met, one point was awarded; if not, no points were awarded. Based on this principle, all training samples were scored as potential labels for each outdoor fitness activity space image of the corresponding type of outdoor fitness space activity.

[0131] In this example, we take basketball (type 5_basketball) as an example: Figure 4 and Figure 5 As shown, two images of outdoor fitness activity spaces are displayed. Figure 4 If an outdoor fitness activity space image satisfies description criteria 1, 2, 3, 4, 6, 7, and 8, then the potential label for this outdoor fitness activity space image is: 7. Figure 5 If an outdoor fitness activity space image satisfies description criteria 1, 2, 3, and 6, then the potential label for this outdoor fitness activity space image is: 4.

[0132] The process of evaluating the activity potential of outdoor fitness spaces based on deep learning is as follows:

[0133] Based on deep learning framework tools, a fully connected image convolutional neural network is built. The feature map sets of each outdoor fitness space activity potential evaluation training set and the potential score label set of the corresponding project are input. Deep learning training is carried out by type, the hyperparameters of the fully connected image convolutional neural network are adjusted, and the multi-classification ability of the fully connected image convolutional neural network for feature images is trained to obtain the multi-classification model of the fully connected image convolutional neural network. Thus, the outdoor fitness space activity potential evaluation model based on deep feature perception is obtained for each type of fitness activity.

[0134] Define an evaluation model for the activity potential of outdoor fitness spaces as follows: , , ,..., , where N is the number of types of outdoor fitness activities.

[0135] The activity potential assessment of the corresponding outdoor fitness space potential points in the urban spatial image is obtained, and the urban spatial image set is used to... Within the activity potential evaluation model for the corresponding outdoor fitness venues, for each image in the urban spatial image set, if it has already been identified as a venue for outdoor fitness activity x during the outdoor fitness space activity type identification process, then in this step, it is further input into the model. Earn activity potential points for outdoor fitness spaces ;

[0136] For the same type of outdoor fitness space activity, K-means cluster analysis was performed on the outdoor fitness space activity potential scores of all urban spatial images, dividing them into levels 3, 2, and 1, representing high potential, medium potential, and low potential, respectively, to obtain the outdoor fitness space activity potential level of the target site. .

[0137] K-means clustering is a typical distance-based clustering algorithm that uses distance as a similarity metric, meaning that the closer two objects are, the greater their similarity. Given a specified number of clusters K, K cases are randomly selected from the dataset as initial cluster centers. The Euclidean distance between the points represented by other cases and the initial cluster centers is calculated, and each case is assigned to the cluster closest to its center. After all data cases are categorized, K datasets (K clusters) are formed. The mean of the data cases in each cluster is recalculated, and this mean is used as the new cluster center. Therefore, the cluster centers are constantly changing, and this process is repeated until convergence, at which point the cluster centers no longer change. In this technical solution, the specified number of clusters k=3, which effectively distinguishes different potential levels, meets application requirements, and reduces data redundancy.

[0138] Based on the geospatial location information of urban spatial images, further spatial analysis can be conducted on the potential level of outdoor fitness activities in the target area.

[0139] In this example, assuming a certain urban spatial image has been identified as a suitable venue for basketball (x=5) and badminton (x=7) activities in the outdoor fitness space activity type identification model, then during the outdoor fitness space activity potential evaluation process, this urban spatial image needs to be input into the respective models. and In the middle, we get:

[0140]

[0141] After K-means clustering classification, the activity potential level of each type of outdoor activity space in the city can be further obtained:

[0142]

[0143] The results of potential evaluation can reflect the specific requirements of different sports and fitness venues for spatial characteristics.

[0144] The results of the potential assessment can be applied in the following scenarios: 1) Analyzing the spatial distribution of existing outdoor fitness spaces in the city and their impact range. In service blind spots, this technology can be used to identify the types and potential of potential outdoor fitness venues, which can efficiently guide the selection of subsequent construction locations and the prioritization of construction, effectively saving resource costs; 2) Based on the composition of residents or fitness needs in different urban blocks, potential spaces can be selectively identified for specific sports; 3) Based on the distribution of specific activity types, urban fitness routes can be planned, and the outdoor fitness space activity potential assessment model can be applied to efficiently focus on areas with low potential and carry out targeted potential enhancement construction.

[0145] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0146] The foregoing has shown and described the basic principles, main features, and advantages of this disclosure. Those skilled in the art should understand that this disclosure is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of this disclosure. Various changes and modifications can be made to this disclosure without departing from its spirit and scope, and all such changes and modifications fall within the scope of this disclosure as claimed.

Claims

1. A method for intelligently identifying the type and potential of outdoor fitness spaces along streets, characterized in that, The method comprises the following steps: acquiring urban space image data sets of a target area; inputting the urban space image data sets of the target area into a pre-established outdoor fitness space activity type identification model, and identifying feature elements in the outdoor fitness space activity type identification system to obtain outdoor fitness space potential points corresponding to the outdoor fitness space activity type of the urban space image; inputting the urban space image data sets of the target area into a pre-established outdoor fitness space activity potential evaluation model corresponding to the outdoor fitness space activity type, and obtaining activity potential evaluation of the outdoor fitness space potential points corresponding to the outdoor fitness space activity type of the urban space image; selecting an outdoor fitness site corresponding to the outdoor fitness space activity type according to the activity potential evaluation of the outdoor fitness space potential points corresponding to the outdoor fitness space activity type in the target area; The outdoor fitness space activity type identification system comprises the following steps: extracting feature elements of the fitness activity space corresponding to each outdoor fitness space activity type according to the feature elements of the fitness activity space corresponding to each outdoor fitness space activity type, and obtaining a feature element set of the fitness activity space corresponding to each outdoor fitness space activity type from the outdoor fitness activity space image data sets; The feature elements of the fitness activity space are divided into decisive positive decision feature elements, general positive decision feature elements and decisive negative decision feature elements according to the degree of fitting the corresponding outdoor fitness space activity type; obtaining the outdoor fitness space potential points corresponding to the outdoor fitness space activity type of the urban space image by identifying the feature elements in the outdoor fitness space activity type identification system, comprising: if the pre-established outdoor fitness space activity type identification model identifies the decisive positive decision elements corresponding to the outdoor fitness activity or only identifies the general positive decision elements, this place is regarded as the outdoor fitness space potential point corresponding to the outdoor fitness space activity type of the urban space image; if the decisive positive decision elements corresponding to the outdoor fitness activity are not identified, and the decisive negative decision elements corresponding to the outdoor fitness activity are identified, or none of the decisive positive decision elements, the decisive negative decision elements and the general positive decision elements corresponding to the outdoor fitness activity are identified, then this place is not regarded as the outdoor fitness space potential point corresponding to the outdoor fitness space activity type of the urban space image; The pre-established outdoor fitness space activity potential evaluation model corresponding to the outdoor fitness space activity type comprises: taking the outdoor fitness space activity potential evaluation system as the evaluation standard, selecting the outdoor fitness activity space image corresponding to the outdoor fitness space activity type in the outdoor fitness activity space image data sets as a training sample for activity potential labeling, and taking it as the outdoor fitness activity activity potential evaluation training set. Based on a deep learning framework tool, a full connection image convolutional neural network is built, an outdoor fitness activity potential evaluation training set of each outdoor fitness activity and a potential score label set of an outdoor fitness activity space image of each corresponding outdoor fitness space activity type are input, a deep learning training is performed on the corresponding outdoor fitness space activity type until the accuracy requirement is met, and an outdoor fitness space activity potential evaluation model of each outdoor fitness space activity type is obtained; The construction of the outdoor fitness space activity potential evaluation system comprises: Each outdoor fitness activity space image of each corresponding outdoor fitness space activity type in the outdoor fitness activity space image data set actually existing the fitness activity is marked for activity potential; The activity potential marking comprises: The number of description points in each outdoor fitness space activity potential evaluation system is summarized, and the corresponding outdoor fitness activity space image is compared one by one according to the description points, and the cumulative addition method is used for scoring, if one description point is met, one point is scored, if not, no points are added, all training samples are scored based on this principle, as the potential marking of the outdoor fitness activity space image of each corresponding outdoor fitness space activity type; The city space image is input into the pre-established outdoor fitness space activity potential evaluation model of the corresponding outdoor fitness space activity type, the outdoor fitness space activity potential score of the corresponding outdoor fitness space activity type is obtained, the outdoor fitness space activity potential scores of the city space images of the target region under the same outdoor fitness space activity type are clustered and graded by K-means, the outdoor fitness space activity potential level is obtained, and the outdoor fitness space activity potential level is used to describe the evaluation result of the outdoor fitness space activity potential of the type. 2.The method of claim 1, wherein, The city space image data set of the target region comprises: The driving track map of the street view collection vehicle of the map open platform in the target region is obtained, the track map is vectorized in the geographic information system platform, the collection interval is determined according to the demand, the vector line data is converted into point data, the city space street view collection point is obtained, the corresponding city space street view collection point is obtained, and the city space image data set is constituted; The city space image includes complete visual image information and geographic space position information. 3.The method of claim 1, wherein, The pre-established outdoor fitness space activity type identification model comprises: Pictures containing feature elements in the feature element set in the existing outdoor fitness space activity type identification system are obtained, the feature element frame selection and type labeling of the pictures containing the feature elements in the feature element set in the existing outdoor fitness space activity type identification system are performed, and the pictures are used as an outdoor fitness space activity type identification training set. The ResNet50 network built based on the deep learning framework tool is used as the bottom network, and the Faster RCNN network is used as the target detection to build an outdoor fitness space type discrimination model. Training samples in the training set of the outdoor fitness space activity type discrimination system are used as input for training, and the feature elements in the training samples are identified until the accuracy of the outdoor fitness space type discrimination model meets the requirements, and the outdoor fitness space activity type discrimination model is established.

4. The method of claim 1, wherein the method further comprises: The positive decision feature elements, general positive decision elements and decisive negative decision feature elements include one or more of the following factors: site size, site shape, site paving, site greenery, fitness facilities and municipal facilities.

5. The method of claim 4, wherein the method further comprises: The actual outdoor fitness activity space image dataset includes: an outdoor fitness actual activity space image dataset corresponding to each outdoor fitness space activity type and an outdoor fitness actual existing space image dataset set. 6.The method of intelligently identifying the type and potential of a street-side outdoor fitness space according to claim 5, wherein, The outdoor fitness actual activity space image dataset corresponding to each outdoor fitness space activity type includes: The geographic space location information is obtained by calling the static panoramic map service of the application programming interface of the map open platform, and the outdoor fitness actual activity space image dataset corresponding to the outdoor fitness space activity type is established. 7.The method of claim 5, wherein the method further comprises: determining the type and potential of the outdoor fitness space based on the obtained information. The outdoor fitness actual existing space image dataset corresponding to each outdoor fitness space activity type includes: an actual existing space street view image set and an actual existing space intercepted image set. The actual existing space street view image set includes: outdoor fitness actual existing space obtained by calling the static panoramic map service of the application programming interface of the map open platform, and the actual existing space street view image set is established. The actual existing space intercepted image set includes: outdoor fitness space that cannot find geographic space location information, and the outdoor fitness actual existing space intercepted image is obtained by selecting a direct intercepted platform image, and the actual existing space intercepted image set is established. The actual existing space intercepted image set includes: outdoor fitness space that cannot find geographic space location information, and the outdoor fitness actual existing space intercepted image is obtained by selecting a direct intercepted platform image, and the actual existing space intercepted image set is established.

Images