Municipal road facility friendliness automatic evaluation method

Abstract

The invention relates to a municipal road facility friendliness automatic evaluation method. The method comprises the following steps: collecting a real-time municipal road facility element image; performing identification and segmentation on the municipal road facility element image by adopting an artificial intelligence image segmentation method, and further calculating a municipal road facility friendliness objective evaluation result; calling a large model to perform subjective evaluation on the municipal road facility element image, and obtaining a municipal road facility friendliness subjective evaluation result in combination with a constructed subjective evaluation model; and combining the municipal road facility friendliness objective evaluation result and the municipal road facility friendliness subjective evaluation result to obtain a municipal road facility friendliness comprehensive evaluation result. Compared with the prior art, the method has the advantages of accurately realizing municipal road facility friendliness evaluation and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of municipal road facility detection and evaluation, and in particular to an automatic evaluation method for the friendliness of municipal road facilities. Background Art

[0002] The friendly design of municipal road facilities is an important factor in improving the travel conditions of urban residents, improving the existing urban ecological environment, and promoting the prosperity and development of cities. It refers to the municipal road design that takes into full consideration the coordination between the road landscape and the natural environment from an aesthetic point of view, so that drivers and passengers feel safe, comfortable and harmonious, and combines the maintenance and management of municipal roads to ensure that the service level of municipal roads remains at a good level. The friendly design of municipal road facilities beautifies engineering protection, makes the toll, refueling and service stations have distinctive styles, beautifies the environment with greening as the main measure, repairs the damage of roads to the natural environment, and increases the cultural connotation of the road environment through the spread of local customs and the embellishment of artificial landscapes along the route, so as to achieve the characteristics of beautiful appearance, strong environmental protection function and strong cultural atmosphere. With the continuous advancement of urbanization, the focus of urban transportation construction is gradually shifting from "construction-oriented" to "construction and maintenance". In this process, residents' requirements for municipal roads are no longer limited to simple connectivity requirements. Furthermore, in addition to connectivity requirements, residents often require municipal roads to be comfortable to drive and municipal road landscape facilities to be pleasant. Therefore, in order to promote the green transformation of road traffic, it is necessary to detect and evaluate the friendliness of municipal road facilities.

[0003] Existing inspection and evaluation methods for municipal roads mostly focus on the practical functions of municipal roads, such as road flatness, road surface disease conditions, road anti-skid ability, etc., but rarely evaluate the friendliness of municipal road facilities, and lack corresponding subjective evaluation methods and indicator systems. Summary of the invention

[0004] The purpose of the present invention is to provide a method for automatically evaluating the friendliness of municipal road facilities, which can accurately realize the automatic evaluation of the friendliness of municipal road facilities.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] A method for automatically evaluating the friendliness of municipal road facilities comprises the following steps:

[0007] Collect real-time images of municipal road facilities elements;

[0008] Using artificial intelligence image segmentation methods to identify and segment the municipal road facility element images, and further calculating objective evaluation results of the friendliness of municipal road facilities;

[0009] Calling the big model to conduct subjective evaluation on the municipal road facility element image, and combining the constructed subjective evaluation model to obtain the subjective evaluation result of the friendliness of the municipal road facilities;

[0010] The objective evaluation result of the friendliness of municipal road facilities and the subjective evaluation result of the friendliness of municipal road facilities are combined to obtain a comprehensive evaluation result of the friendliness of municipal road facilities.

[0011] Furthermore, the method also includes performing data enhancement processing on the municipal road facility element image before segmentation, wherein the data enhancement processing operations include mirroring, cropping, adding noise and brightness transformation.

[0012] Furthermore, the step of calculating the objective evaluation result of the friendliness of municipal road facilities includes:

[0013] Based on the artificial intelligence image segmentation method, the municipal road facility element image is identified and segmented to obtain the area proportion of each type of facility element, and the municipal road pavement is identified and segmented to obtain the municipal road pavement condition;

[0014] The weights of the friendliness of the various types of facility elements to the municipal roads are set, and the landscape area ratio LAR on the municipal road facility element image is calculated in combination with the area proportions of the various types of facility elements, where the calculation expression is:

[0015]

[0016] In the formula, w i is the set weight, A i is the area proportion of the i-th type of facility elements, and A is the total pixel area of ​​the municipal road facility element image;

[0017] According to the municipal road pavement conditions, the number of pavement defects is counted D 1 and the number of foreign objects on the road 2 , calculate the municipal road condition index RCI, where the calculation expression is:

[0018] RCI=5×∑D i

[0019] Where D i is the number of pavement defects or foreign objects on the pavement, i = 1, 2;

[0020] Based on the landscape area ratio LAR and the municipal road condition index RCI, the objective evaluation result of the friendliness of municipal road facilities is calculated, wherein the calculation expression is:

[0021] OURFI=LAR-RCI

[0022] Where OURFI is an objective evaluation index of the friendliness of municipal road facilities.

[0023] Furthermore, the various types of facility elements include vegetation, sky, roads, buildings and vehicle areas.

[0024] Furthermore, the artificial intelligence image segmentation method uses a fine-tuned semantic segmentation model for recognition and segmentation, and the fine-tuning step of the semantic segmentation model includes:

[0025] Obtain a collection of municipal road facility element images and perform data enhancement processing;

[0026] The image annotation tool Labelme is used to perform pixel-level semantic segmentation and annotation of various types of facility elements in each municipal road facility element image after data enhancement, and the municipal road pavement diseases and foreign objects are also annotated to obtain the municipal road facility element segmentation dataset;

[0027] The semantic segmentation model is fine-tuned based on the municipal road facility element segmentation dataset to obtain a fine-tuned semantic segmentation model, wherein the image encoding layer and the prompt word encoding layer in the semantic segmentation model are in a frozen state during the fine-tuning process.

[0028] Furthermore, the step of obtaining the subjective evaluation result of the friendliness of municipal road facilities includes:

[0029] Based on the municipal road facility element image, multiple large models are called to score each influencing factor and calculate the average score of each influencing factor, wherein the influencing factors include roadside greening design, visibility and light intensity, road sign design, road surface cleanliness and road damage condition;

[0030] The average score of each influencing factor is input into the constructed subjective evaluation model, and the subjective evaluation result of the friendliness of municipal road facilities is output.

[0031] Furthermore, the calculation expression of the subjective evaluation result of the friendliness of municipal road facilities is:

[0032] SURFI=y s =w s,1 x s,1 +w s,2 x s,2 +...+w s,n x s，n

[0033] In the formula, SURFI, y s is the subjective evaluation result score of the friendliness of municipal road facilities, w s,n is the weight of the nth influencing factor, x s,n is the average score of the nth influencing factor.

[0034] Furthermore, the steps of constructing the subjective evaluation model include:

[0035] Setting the hierarchy of the subjective evaluation model, including a target layer for scoring the subjective evaluation results of the friendliness of municipal road facilities and a criterion layer for scoring each influencing factor in the element image of the municipal road facilities;

[0036] The influencing factors in the criterion layer were compared based on the opinions of experts and residents to obtain relative importance, and the relative importance scale of each influencing factor was obtained by combining the Santy scaling method and filled in the judgment matrix of the relative importance scale;

[0037] Based on the judgment matrix of the relative importance scale, the weight of each influencing factor is calculated to complete the construction of the criterion layer, wherein the calculation expression of the weight is:

[0038]

[0039] In the formula, w i is the weight of the i-th influencing factor, n is the number of influencing factors, a ij 、a kj are the elements in the i-th row and j-th column and the elements in the k-th row and j-th column in the judgment matrix of the relative importance scale, respectively;

[0040] According to the constructed criterion layer, the target layer is constructed to form a subjective evaluation model.

[0041] Furthermore, the step of obtaining the comprehensive evaluation result of the friendliness of municipal road facilities includes:

[0042] The analytic hierarchy process is used to determine the subjective weights w of the objective evaluation results of the friendliness of municipal road facilities and the subjective evaluation results of the friendliness of municipal road facilities. so and w ss ;

[0043] The entropy weight method is used to determine the objective weights w of the objective evaluation results of the friendliness of municipal road facilities and the subjective evaluation results of the friendliness of municipal road facilities. oo and w os ;

[0044] Based on the subjective weight and the objective weight, a combined weight is calculated, wherein the calculation expression of the combined weight is:

[0045]

[0046] Where W o is the combined weight of the objective evaluation results of the friendliness of municipal road facilities, W sis the combined weight of the subjective evaluation results of the friendliness of municipal road facilities, w oo 、w os is the objective weight of the objective evaluation result of the friendliness of municipal road facilities and the subjective evaluation result of the friendliness of municipal road facilities, w so 、w ss The subjective weights of the objective evaluation results and the subjective evaluation results of the friendliness of municipal road facilities;

[0047] Based on the combined weight W o and W s , combining the objective evaluation result OURFI of the friendliness of municipal road facilities and the subjective evaluation result SURFI of the friendliness of municipal road facilities, the comprehensive evaluation result URFI of the friendliness of municipal road facilities is obtained, wherein the calculation expression of the comprehensive evaluation result of the friendliness of municipal road facilities is:

[0048] URFI=W o OURFI+W s SURFI

[0049] Where URFI is the comprehensive evaluation result of the friendliness of municipal road facilities, OURFI is the objective evaluation result of the friendliness of municipal road facilities, and SURFI is the subjective evaluation result of the friendliness of municipal road facilities.

[0050] Furthermore, the entropy weight method is used to determine the objective weight w oo and w os The steps include:

[0051] Based on the objective evaluation results of the friendliness of municipal road facilities and the subjective evaluation results of the friendliness of municipal road facilities, normalization is performed, wherein the normalized expression is:

[0052]

[0053] In the formula, x is the objective evaluation index of the friendliness of municipal road facilities or the subjective evaluation index of the friendliness of municipal road facilities, min(x) and max(x) are the minimum and maximum indexes respectively;

[0054] According to the normalization results, the information entropy E of the objective evaluation index of municipal road facility friendliness and the subjective evaluation index of municipal road facility friendliness are calculated. j , where information entropy E j The calculation expression is:

[0055]

[0056] In the formula, n is the number of samples involved in calculating information entropy, p ij is the numerical weight of the i-th sample of the j-th indicator, xij is the normalized value of the i-th sample of the j-th indicator;

[0057] Based on the information entropy E of each indicator j The objective weight of each indicator is calculated to obtain the objective weight of the objective evaluation result of the friendliness of municipal road facilities and the subjective evaluation result of the friendliness of municipal road facilities, wherein the calculation expression of the objective weight is:

[0058]

[0059] In the formula, w j is the objective weight of the jth indicator for the comprehensive evaluation result of the friendliness of municipal road facilities, k is the number of indicators involved in the calculation of the objective weight, here k = 2.

[0060] Compared with the prior art, the present invention has the following beneficial effects:

[0061] (1) For municipal road facility element images, the present invention adopts an artificial intelligence image segmentation method to objectively evaluate the friendliness of municipal road facilities and calls a large model combined with a subjective evaluation model to subjectively evaluate the friendliness of municipal road facilities. The objective evaluation and the subjective evaluation are combined to perform a comprehensive evaluation, thereby accurately realizing the automatic evaluation of the friendliness of municipal road facilities.

[0062] (2) The semantic segmentation model adopted by the present invention can not only identify and segment various types of facility elements, but also identify and segment municipal road pavement defects and foreign objects, thereby considering the impact of pavement conditions and various types of facility elements on friendliness in the objective evaluation process and improving the accuracy of objective evaluation.

[0063] (3) The subjective evaluation model of the present invention takes into account a variety of influencing factors for subjective evaluation. By calculating the weight of each influencing factor and combining the average score of each influencing factor with the large model, the accuracy of the subjective evaluation can be improved.

[0064] (4) The semantic segmentation model adopted by the present invention can be used only by fine-tuning, and subjective evaluation can be performed by calling multiple existing large models, saving computing power and computing costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0066] Figure 2 Schematic diagrams of the installation positions of the industrial camera of the present invention, wherein (a) is a schematic diagram of the installation position on a motor vehicle, and (b) is a schematic diagram of the installation position on a non-motor vehicle;

[0067] Figure 3This is a diagram of the artificial intelligence image segmentation algorithm architecture of the present invention;

[0068] Figure 4 This is a diagram of the fine-tuned artificial intelligence image segmentation algorithm architecture of the present invention;

[0069] Figure 5 The road facility element recognition result diagram of the present invention, wherein (a) is a collected motor vehicle lane facility element image, and (b) is a motor vehicle lane facility element recognition result;

[0070] Figure 6 The pavement disease and foreign body recognition result diagram of the present invention, wherein (a) is the collected motor vehicle lane facility element image, and (b) is the motor vehicle road pavement condition recognition result;

[0071] Figure 7 This is a diagram showing the subjective evaluation results of the large model of the present invention on the road friendliness. DETAILED DESCRIPTION

[0072] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0073] This embodiment provides an automatic evaluation method for the friendliness of municipal road facilities. The method uses an industrial camera to collect image data of municipal road facilities, and combines an artificial intelligence image segmentation algorithm to count the road facility elements in the image, and then proposes an objective evaluation index for the friendliness of road facilities for objective evaluation; on the other hand, a subjective evaluation system for the friendliness of road facilities is established using the hierarchical analysis method, and a large model is called as an expert in the subjective evaluation of the friendliness of road facilities to subjectively score the friendliness of municipal road facilities. Finally, a combined weighting method is used to combine objective evaluation and subjective evaluation, to establish a municipal road facility friendliness evaluation system that combines subjective and objective aspects, and to achieve automatic detection and evaluation of the friendliness of municipal road facilities. Figure 1 As shown, the method comprises the following steps:

[0074] The first step is equipment placement and image acquisition.

[0075] In order to evaluate the friendliness of municipal road facilities, it is necessary to first collect information on the status of municipal road facility elements. The present invention installs an industrial camera on a detection vehicle, and completes the collection of road facility element information by driving the detection vehicle equipped with the industrial camera on the road. Figure 2 As shown, they can be installed in motor vehicles ( Figure 2 (a) in the figure) or non-motor vehicles ( Figure 2The information of road facilities observed by non-motor vehicles is collected, and the installation perspective should simulate the vision of residents when driving as much as possible, so as to give a more scientific and reasonable evaluation.

[0076] The second step is objective evaluation of image segmentation algorithm.

[0077] The objective analysis of the friendliness of the collected road facility element images is to analyze the size and proportion of blocks belonging to different types in the road facility element images, and identify and detect the road conditions. For example, when there are more sky and vegetation in the road facility element images, it is a more friendly road facility; when there are more high-rise buildings, roads and vehicles in the road facility element images, the friendliness of the road facilities is poor; on the other hand, if there are road surface diseases such as cracks and potholes in the road facility element images, the friendliness of the road facilities is also poor. Therefore, it is necessary to automatically distinguish the facility element areas on the image and segment the blocks with different attributes in the image. In the computer, the image is stored in the form of pixels, and a pixel is often represented by 3 numbers from 0 to 255, representing the depth of the three colors of red (R), green (G), and blue (B), and these three colors constitute the various colors in the image. Therefore, the friendliness analysis of the road facilities of the image is to find the pixels belonging to each element category in the image and perform statistical analysis. This process can usually be performed using semantic segmentation. Semantic segmentation is a basic task in computer vision. It refers to the segmentation and classification of images at the pixel level, so that each pixel obtains a corresponding semantic label. Based on deep learning neural networks, computers can autonomously learn and analyze pixels with the same semantic features from a large number of labeled data sets. With the help of deep learning semantic segmentation algorithms, computers can autonomously find the features of the target, thereby achieving the target recognition task.

[0078] In semantic segmentation, Facebook's Meta AI Lab has built a semantic segmentation model, Segment Anything Model (SAM), which has learned the general concept of objects by learning from more than 11 million images and can be used for preliminary recognition and segmentation of any image or video. Its algorithm architecture is as follows: Figure 3 As shown in the figure, it mainly includes three parts: image encoding, prompt word encoding and mask decoding. Among them, image encoding acts on the input image, and summarizes the characteristics of the image by extracting the high-dimensional features and semantic information of the image; prompt word encoding acts on the input text, points, rectangular boxes and masks, etc., and converts them into information that the computer can understand for processing; finally, by combining image feature information and input prompts, the algorithm model can autonomously segment and identify target objects in the image with high precision and output reliable results.

[0079] On this basis, the present invention uses the road facility element image acquisition platform established in the first step to collect a large number of road facility element images, selects representative road facility element images, and performs data enhancement including mirroring, cropping, adding noise and brightness transformation. Then, the open source image annotation tool Labelme is used to perform pixel-level semantic segmentation and annotation of various elements in each road facility element image after data enhancement. The annotated elements include vegetation, sky, road, building, vehicle, etc., and road surface diseases and foreign objects are annotated at the same time, so as to establish a road facility element segmentation data set.

[0080] Furthermore, the present invention freezes the image encoder and prompt encoder in SAM, such as Figure 4 As shown in the figure. The SAM model is fine-tuned using the established road facility element segmentation dataset, and the SAM model is fine-tuned from a general image segmentation model to a dedicated model specializing in road facility element image segmentation. The acquired road facility element image is identified and segmented using the dedicated model for road facility element image segmentation. The results are shown in the figure. Figure 5 As shown in (a) and (b) of the figure, the road condition is identified and segmented, and the results are as follows Figure 6 As shown in Figure (a) and Figure (b).

[0081] In addition, embodiments of the present invention may also apply other deep learning algorithms for image segmentation, including but not limited to fully connected neural network FCN, symmetric semantic segmentation model UNet, deep convolutional semantic segmentation model DeepLab, etc.; when conditions permit, it is advisable to adopt large visual models, including but not limited to segmentation everything model (Segment Anything Model), general segmentation model (SegGPT), etc.

[0082] exist Figure 5 Based on this, the features are further divided to extract the areas belonging to vegetation, sky, road, building and vehicle, and their pixel areas A are counted respectively. i , and the total pixel area A of the image. Then, the weights w of vegetation, sky, road, building, and vehicle areas on the friendliness of municipal roads are set i On this basis, the landscape area ratio LAR of the image is calculated:

[0083]

[0084] exist Figure 6 Based on the number of road surface defects, the number of road surface defects D 1 and the number of foreign objects on the road 2, calculate the municipal road condition index RCI:

[0085] RCI=5×∑D i

[0086] The objective index of friendliness of municipal road facilities, OURFI, is calculated as follows:

[0087] OURFI=LAR-RCI

[0088] The third step is subjective evaluation of the large model.

[0089] The analytic hierarchy process (AHP) is a subjective value assignment evaluation method. It decomposes the elements related to decision-making into multiple levels such as the target layer, decision layer, and program layer, and then judges the relative importance between the two elements and ranks them according to the weights calculated by importance. On this basis, qualitative and quantitative analysis is performed to obtain the final result. In the subjective evaluation model established in this paper, the target layer is the subjective evaluation result score of the friendliness of municipal road facilities, and the criterion layer is the score of the five factors of roadside greening design, visibility and light intensity, road sign design, road surface cleanliness, and road damage. The relative importance of the five factors in the criterion layer is compared by referring to the opinions of experts and residents, and the specific relative importance scale is determined according to the Santy scale method shown in Table 1 and filled in the judgment matrix.

[0090] Table 1 Relative importance scale

[0091]

[0092] According to the judgment matrix, the weight parameters of each factor in the criterion layer can be calculated using the following formula:

[0093]

[0094] Based on this, a subjective evaluation model is established:

[0095] SURFI=y s =w s，1 x s，1 +w s，2 x s，2 +...+w s，n x s，n

[0096] After the subjective evaluation model of the friendliness of municipal road facilities is established, the five aspects of roadside greening design, visibility and light intensity, road sign design, road surface cleanliness and road damage of the road facilities can be scored according to the collected road facility element images, and the subjective evaluation result score of the friendliness of municipal road facilities can be obtained. To this end, the present invention acts as an expert judge by calling the big model development interface API, and uses the big model to identify and analyze the road facility element images, complete the scoring of various aspects of the road facility element images, and thus realize the subjective evaluation of the friendliness of road facilities. The big models called include ChatGPT-4o, BERT, LLaMA, Wenxin Yiyan, Tongyi Qianwen, Zhipu Qingyan ChatGLM, etc. The submitted content includes:

[0097] (1).[Road facility element image]

[0098] (2) Score the road from five perspectives: roadside greening design, visibility and light intensity, road sign design, road surface cleanliness, and road damage. The score is between 0 and 10, and explain the reasons.

[0099] After obtaining the subjective evaluation scores of the road facility element images from each large model, the average values ​​of roadside greening design, visibility and light intensity, road sign design, road surface cleanliness and road damage condition were taken as input into the subjective evaluation model of municipal road facility friendliness to obtain the subjective evaluation result score SURFI of municipal road facility friendliness.

[0100] SURFI=y s =w s,1 x s,1 +w s，2 x s，2 +...+w s，n x s，n

[0101] The fourth step is comprehensive evaluation using the combined weighting method.

[0102] The combined weighting method is a weighting method that takes both subjective factors and objective indicators into consideration. It can make up for the shortcomings of single weighting, realize the internal unity of subjective and objective factors, and make the evaluation results real and scientific. In terms of specific implementation, the combined weighting method first uses the hierarchical analysis method to obtain the subjective weights of the subjective evaluation and objective evaluation of the friendliness of municipal road facilities. oo 、w os , and then use the entropy weight method to obtain the objective weights w of subjective evaluation and objective evaluation respectively so 、w ss , and finally determine the combined weight.

[0103] Firstly, the analytic hierarchy process is used to determine the subjective weights w of the subjective and objective evaluation results. oo 、w os According to the feedback from experts, the subjective evaluation results are more comprehensive and therefore slightly more important than the objective evaluation results. Therefore, the judgment matrix is ​​shown in Table 2 below:

[0104] Table 2 Judgment Matrix

[0105]

[0106] The weight of the subjective evaluation result is calculated as: w oo =0.67; the weight of objective evaluation result is w os =0.33.

[0107] Then the entropy weight method is used to determine the objective weights w of the subjective evaluation results and the objective evaluation results. so 、w ss The entropy weight method is an objective weighting method. It starts from the data itself, calculates the entropy weight of each indicator according to the degree of variation of each indicator, and then corrects the weight of each indicator by entropy weight, so as to obtain objective indicator weight. When using the entropy weight method to weight indicators and establish an evaluation system, the data must first be normalized to eliminate the influence of the dimension. The normalization formula is shown as follows:

[0108]

[0109] Then calculate the information entropy E of each indicator j :

[0110]

[0111] Finally, the objective weight w of each indicator is obtained according to the following formula: j :

[0112]

[0113] Thus, the objective weights w of the subjective evaluation results and the objective evaluation results are obtained. so 、w ss .

[0114] Finally, the combined weight of the subjective evaluation results and the objective evaluation results is calculated according to the following formula:

[0115]

[0116] Based on this, the comprehensive evaluation results of the friendliness of municipal road facilities URFI are obtained:

[0117] URFI=W o OURFI+W sSURFI

[0118] The present invention takes a municipal road facility at a certain place as an example for verification, and the specific steps include:

[0119] (1) Using an industrial camera mounted on a detection vehicle to collect images of municipal road facilities

[0120] In order to collect images of municipal road facility elements, the present invention installs industrial cameras on motor vehicles and bicycles respectively, and uses the movement of the inspection vehicle to collect a large number of municipal road facility element images around the Jiading campus of Tongji University for subsequent analysis.

[0121] (2) Objectively analyze and evaluate road facility element images using image segmentation algorithms

[0122] The SAM segmentation model is used to segment the collected municipal road facility element images, and the areas of vegetation, sky, road, building and vehicle areas are counted as: 133756, 122753, 84739, 0, 0, respectively. The total pixel area of ​​the original image is: 480000. For the weight parameters of various facility elements, the road area is the basic element, so it is taken as 1; the vegetation area and sky area are positive elements, so they are taken as 1.5; buildings and vehicles are negative elements, so they are taken as -1, thus calculating:

[0123]

[0124] On the other hand, it is recognized that there is a crack in the road facility element image, and there is no foreign matter on the road, so the calculation is:

[0125] RCI=5×1=5

[0126] Therefore, the objective evaluation index of the friendliness of municipal road facilities, OURFI, is:

[0127] OLFI=LAR-RCI=97.813-5=92.813

[0128] That is to say, the objective evaluation result of the friendliness of the municipal road facilities represented by the image is 92.813.

[0129] (3) Subjective analysis and evaluation of road facility element images using the analytic hierarchy process and large models

[0130] In view of the five factors that affect the friendliness of municipal road facilities: roadside greening design, visibility and light intensity, road sign design, road surface cleanliness and road damage, with reference to the opinions of experts and residents, and combined with the Santy scale matrix, a judgment matrix is ​​listed as shown in Table 3:

[0131] Table 3 Subjective judgment matrix

[0132]

[0133] The weight parameters of each factor in the criterion layer are calculated as follows: 1 =0.102, w 2 =0.253, w 3 =0.053, w 4 =0.102, w 5 =0.490. Therefore, the expression of the subjective evaluation model of the friendliness of municipal road facilities is:

[0134] SURFI=y=0.102x 1 +0.253x 2 +0.053x 3 +0.102x 4 +0.490x 5

[0135] According to the subjective evaluation model, the images of municipal road facilities collected around the Jiading campus of Tongji University were uploaded to the big model, and the scores of the five factors of roadside greening design, visibility and light intensity, road sign design, road surface cleanliness and road damage were obtained as follows: 8.6, 8.4, 9.5, 9.3, 9.2. The subjective score of the friendliness of the municipal road facilities represented by the image was calculated as follows: SURFI = y = 8.9625

[0136] (4) Comprehensive evaluation of municipal road environment landscape using combined weighting method

[0137] First, the judgment matrix of the subjective evaluation results is listed using the hierarchical analysis method. According to the feedback from experts, the subjective evaluation results are more comprehensive and therefore slightly more important than the objective evaluation results. Therefore, the judgment matrix is ​​shown in Table 4 below:

[0138] Table 4 Evaluation matrix

[0139]

[0140] The weight of the subjective evaluation result is calculated as: w oo =0.67; the weight of objective evaluation result is: w os =0.33.

[0141] Secondly, according to the previous method, the five roads around the Jiading campus of Tongji University were evaluated subjectively and objectively. Since the range of subjective evaluation is 0-10 and the range of objective evaluation is 0-100, in order to balance the ranges of the two, the subjective evaluation results are magnified 10 times. The evaluation results are shown in Table 5 below:

[0142] Table 5 Evaluation results

[0143]

[0144] Based on this, the entropy weight method is used to calculate the weights of the two variables: the weight of the subjective evaluation result w so =0.533; the objective evaluation result weight is w ss =0.467.

[0145] Therefore, the weight calculated by the combined weighting method is:

[0146]

[0147] That is, the evaluation model combining subjective and objective factors for the friendliness of municipal road facilities is:

[0148] URFI=0.604OURFI+0.396SURFI

[0149] Therefore, a comprehensive evaluation combining subjective and objective factors was conducted on the above five roads, and the following results were obtained: Figure 7 And the results in Table 6 below:

[0150] Table 6 Comprehensive evaluation results

[0151]

[0152] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0153] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes. The schemes in the embodiments of the present invention may be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.

[0154] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0155] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0156] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0157] Although the preferred embodiments of the present invention have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0158] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for automatically evaluating the friendliness of municipal road facilities, characterized in that: The following steps are involved: Collect real-time images of municipal road facilities elements; Using artificial intelligence image segmentation methods to identify and segment the municipal road facility element images, and further calculating objective evaluation results of the friendliness of municipal road facilities; Calling the big model to conduct subjective evaluation on the municipal road facility element image, and combining the constructed subjective evaluation model to obtain the subjective evaluation result of the friendliness of the municipal road facilities; The objective evaluation result of the friendliness of municipal road facilities and the subjective evaluation result of the friendliness of municipal road facilities are combined to obtain a comprehensive evaluation result of the friendliness of municipal road facilities.

2. The method for automatically evaluating the friendliness of municipal road facilities according to claim 1, characterized in that: The method also includes performing data enhancement processing on the municipal road facility element image before segmentation, wherein the data enhancement processing operations include mirroring, cropping, adding noise and brightness transformation.

3. The method for automatically evaluating the friendliness of municipal road facilities according to claim 1, characterized in that: The step of calculating the objective evaluation result of the friendliness of municipal road facilities comprises: Based on the artificial intelligence image segmentation method, the municipal road facility element image is identified and segmented to obtain the area proportion of each type of facility element, and the municipal road pavement is identified and segmented to obtain the municipal road pavement condition; The weights of the friendliness of the various types of facility elements to the municipal roads are set, and the landscape area ratio LAR on the municipal road facility element image is calculated in combination with the area proportions of the various types of facility elements, where the calculation expression is: In the formula, w i is the set weight, A i is the area proportion of the i-th type of facility elements, and A is the total pixel area of ​​the municipal road facility element image; According to the municipal road pavement condition, the number of pavement diseases D1 and the number of pavement foreign objects D2 are counted to calculate the municipal road condition index RCI, where the calculation expression is: RCI=5×∑D i Where D i is the number of pavement defects or foreign objects on the pavement, i = 1, 2; Based on the landscape area ratio LAR and the municipal road condition index RCI, the objective evaluation result of the friendliness of municipal road facilities is calculated, wherein the calculation expression is: OURFI=LAR-RCI Where OURFI is an objective evaluation index of the friendliness of municipal road facilities.

4. The method for automatically evaluating the friendliness of municipal road facilities according to claim 3 is characterized in that: The various types of facility elements include vegetation, sky, roads, buildings, and vehicle areas.

5. The method for automatically evaluating the friendliness of municipal road facilities according to claim 1, characterized in that: The artificial intelligence image segmentation method uses a fine-tuned semantic segmentation model for recognition and segmentation, and the fine-tuning steps of the semantic segmentation model include: Obtain a collection of municipal road facility element images and perform data enhancement processing; The image annotation tool Labelme is used to perform pixel-level semantic segmentation and annotation of various types of facility elements in each municipal road facility element image after data enhancement, and the municipal road pavement diseases and foreign objects are also annotated to obtain the municipal road facility element segmentation dataset; The semantic segmentation model is fine-tuned based on the municipal road facility element segmentation dataset to obtain a fine-tuned semantic segmentation model, wherein the image encoding layer and the prompt word encoding layer in the semantic segmentation model are in a frozen state during the fine-tuning process.

6. The method for automatically evaluating the friendliness of municipal road facilities according to claim 1, characterized in that: The steps of obtaining the subjective evaluation results of the friendliness of municipal road facilities include: Based on the municipal road facility element image, multiple large models are called to score each influencing factor and calculate the average score of each influencing factor, wherein the influencing factors include roadside greening design, visibility and light intensity, road sign design, road surface cleanliness and road damage condition; The average score of each influencing factor is input into the constructed subjective evaluation model, and the subjective evaluation result of the friendliness of municipal road facilities is output.

7. The method for automatically evaluating the friendliness of municipal road facilities according to claim 6, characterized in that: The calculation expression of the subjective evaluation result of the friendliness of municipal road facilities is: SURFI=y s =w s，1 x s,1 +w s,2 x s,2 +…+w s，n x s，n In the formula, SURFI, y s is the subjective evaluation result score of the friendliness of municipal road facilities, w s,n is the weight of the nth influencing factor, x s,n is the average score of the nth influencing factor.

8. The method for automatically evaluating the friendliness of municipal road facilities according to claim 1, characterized in that: The steps of constructing the subjective evaluation model include: Setting the hierarchy of the subjective evaluation model, including a target layer for scoring the subjective evaluation results of the friendliness of municipal road facilities and a criterion layer for scoring each influencing factor in the element image of the municipal road facilities; The influencing factors in the criterion layer were compared based on the opinions of experts and residents to obtain relative importance, and the relative importance scale of each influencing factor was obtained by combining the Santy scaling method and filled in the judgment matrix of the relative importance scale; Based on the judgment matrix of the relative importance scale, the weight of each influencing factor is calculated to complete the construction of the criterion layer, wherein the calculation expression of the weight is: In the formula, w i is the weight of the i-th influencing factor, n is the number of influencing factors, a ij 、a kj are the elements in the i-th row and j-th column and the elements in the k-th row and j-th column in the judgment matrix of the relative importance scale, respectively; According to the constructed criterion layer, the target layer is constructed to form a subjective evaluation model.

9. The method for automatically evaluating the friendliness of municipal road facilities according to claim 1, characterized in that: The steps of obtaining the comprehensive evaluation results of the friendliness of municipal road facilities include: The analytic hierarchy process is used to determine the subjective weights w of the objective evaluation results of the friendliness of municipal road facilities and the subjective evaluation results of the friendliness of municipal road facilities. so and w ss ; The entropy weight method is used to determine the objective weights w of the objective evaluation results of the friendliness of municipal road facilities and the subjective evaluation results of the friendliness of municipal road facilities. oo and w os ; Based on the subjective weight and the objective weight, a combined weight is calculated, wherein the calculation expression of the combined weight is: Where W o is the combined weight of the objective evaluation results of the friendliness of municipal road facilities, W s is the combined weight of the subjective evaluation results of the friendliness of municipal road facilities, w oo 、w os is the objective weight of the objective evaluation result of the friendliness of municipal road facilities and the subjective evaluation result of the friendliness of municipal road facilities, w so 、w ss The subjective weights of the objective evaluation results and the subjective evaluation results of the friendliness of municipal road facilities; Based on the combined weight W o and W s , combining the objective evaluation result OURFI of the friendliness of municipal road facilities and the subjective evaluation result SURFI of the friendliness of municipal road facilities, the comprehensive evaluation result URFI of the friendliness of municipal road facilities is obtained, wherein the calculation expression of the comprehensive evaluation result of the friendliness of municipal road facilities is: <h2 style=";text-align:left;direction:ltr">URFI=W<h2 style=";text-align:left;direction:ltr"> o <h2 style=";text-align:left;direction:ltr"> OURFI+W<h2 style=";text-align:left;direction:ltr"> s <h2 style=";text-align:left;direction:ltr"> SURFI Where URFI is the comprehensive evaluation result of the friendliness of municipal road facilities, OURFI is the objective evaluation result of the friendliness of municipal road facilities, and SURFI is the subjective evaluation result of the friendliness of municipal road facilities.

10. The method for automatically evaluating the friendliness of municipal road facilities according to claim 9, characterized in that: The entropy weight method is used to determine the objective weight w oo and w os The steps include: Based on the objective evaluation results of the friendliness of municipal road facilities and the subjective evaluation results of the friendliness of municipal road facilities, normalization is performed, wherein the normalized expression is: In the formula, x is the objective evaluation index of the friendliness of municipal road facilities or the subjective evaluation index of the friendliness of municipal road facilities, min(x) and max(x) are the minimum and maximum indexes respectively; According to the normalized results, the information entropy E of the objective evaluation index of municipal road facility friendliness and the subjective evaluation index of municipal road facility friendliness are calculated respectively. j , where information entropy E j The calculation expression is: In the formula, n is the number of samples involved in calculating information entropy, p ij is the numerical weight of the i-th sample of the j-th indicator, x ij is the normalized value of the i-th sample of the j-th indicator; Based on the information entropy E of each indicator j The objective weight of each indicator is calculated to obtain the objective weight of the objective evaluation result of the friendliness of municipal road facilities and the subjective evaluation result of the friendliness of municipal road facilities, wherein the calculation expression of the objective weight is: In the formula, w j is the objective weight of the jth indicator for the comprehensive evaluation result of the friendliness of municipal road facilities, k is the number of indicators involved in the calculation of the objective weight, here k = 2.

Images