Digital garden landscape management system and method based on visualization technology

Abstract

The invention relates to the field of digital garden landscape management, and particularly discloses a digital garden landscape management system and method based on a visualization technology, and the method comprises the steps: firstly obtaining a high-resolution green land coverage image shot by an unmanned plane and garden landscape environment data collected by a sensor, and then carrying out the deep learning technology, and feature extraction and correlation analysis are carried out on the two images, and finally, a visual analysis report of the health state of the landscape architecture is generated through a generator, so that the health state and potential risks of the landscape architecture are known, and effective management and maintenance measures are taken.

Description

Technical Field

[0001] The present application relates to the field of digital garden landscape management, and more specifically, to a digital garden landscape management system and method based on visualization technology. Background Art

[0002] With the acceleration of urbanization, the area and density of green spaces in cities have gradually increased, and landscaping has become an important part of the modern urban environment. Urban gardens refer to various green spaces and gardening designs in cities, including parks, gardens, square green spaces, and street green belts. These green spaces not only enhance the beauty of the city, but also provide a place for citizens to relax, socialize, and have fun.

[0003] With the development of cities and the increase in the number of green spaces, the scale and layout of landscaping have become increasingly complex. Currently, most landscaping management still relies on manual inspection and maintenance, which is not only inefficient but also limited by human resources and management costs. Traditional management methods are difficult to provide comprehensive and accurate garden data, resulting in green space maintenance and management often relying on experience and rough estimates.

[0004] Therefore, a digital garden landscape management system and method based on visualization technology is desired. Summary of the invention

[0005] In order to solve the above technical problems, this application is proposed. The embodiment of this application provides a digital garden landscape management system and method based on visualization technology, which first obtains high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors, then uses deep learning technology to perform feature extraction and correlation analysis on the two, and finally generates a garden landscape health status visualization analysis report through a generator to understand the health status and potential risks of the garden, and then take effective management and maintenance measures.

[0006] According to one aspect of the present application, a digital garden landscape management system based on visualization technology is provided, which includes:

[0007] Digital garden landscape data acquisition module, used to obtain high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors;

[0008] A digital garden landscape data extraction module is used to extract high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor;

[0009] The landscape health status report generation module is used to generate a visual analysis report on the garden landscape health status based on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector.

[0010] According to another aspect of the present application, a digital garden landscape management method based on visualization technology is provided, which includes:

[0011] Obtain high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors;

[0012] Extracting high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor;

[0013] Based on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector, a garden landscape health status visualization analysis report is generated.

[0014] Compared with the existing technology, the present application provides a digital garden landscape management system and method based on visualization technology, which first obtains high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors, and then uses deep learning technology to perform feature extraction and correlation analysis on the two. Finally, a generator is used to generate a visualization analysis report on the health status of the garden landscape, so as to understand the health status and potential risks of the garden, and then take effective management and maintenance measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0016] Figure 1 It is a block diagram of a digital garden landscape management system based on visualization technology according to an embodiment of the present application.

[0017] Figure 2 It is a block diagram of a digital garden landscape data extraction module in a digital garden landscape management system based on visualization technology according to an embodiment of the present application.

[0018] Figure 3 It is a block diagram of a green space coverage image feature extraction unit in a digital garden landscape management system based on visualization technology according to an embodiment of the present application.

[0019] Figure 4 It is a block diagram of a garden landscape environment data feature extraction unit in a digital garden landscape management system based on visualization technology according to an embodiment of the present application.

[0020] Figure 5 The block diagram is a schematic diagram of a landscape health status report generation module in a digital garden landscape management system based on visualization technology according to an embodiment of the present application.

[0021] Figure 6 The present invention is a flowchart of a digital garden landscape management method based on visualization technology according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] Various exemplary embodiments, features and aspects of the present application will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0023] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0024] In addition, in order to better illustrate the present application, numerous specific details are given in the following specific embodiments. It should be understood by those skilled in the art that the present application can also be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present application.

[0025] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0026] Figure 1 Schematic diagram of a digital garden landscape management system based on visualization technology according to an embodiment of the present application. Figure 1As shown, according to the digital garden landscape management system 100 based on visualization technology according to the embodiment of the present application, it includes: a digital garden landscape data acquisition module 110, which is used to acquire high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors; a digital garden landscape data extraction module 120, which is used to extract high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage images taken by drones and the garden landscape environment data collected by sensors; a landscape health status report generation module 130, which is used to generate a garden landscape health status visualization analysis report based on the high-resolution green space coverage global correlation feature vectors and the garden landscape environment global correlation feature vectors.

[0027] In the above-mentioned digital garden landscape management system 100 based on visualization technology, the digital garden landscape data acquisition module 110 is used to obtain high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors. It should be understood that with the rapid advancement of urbanization, the area and density of green spaces in cities are constantly expanding, and garden greening has become a key component of the modern urban environment. Urban gardens include various green spaces and gardening designs, such as parks, gardens, square green spaces, and street green belts. These green spaces not only enhance the visual beauty of the city, but also provide residents with space for relaxation, socializing and leisure. However, with the development of cities and the increase in the number of green spaces, the scale and layout of garden greening have become more and more complex. At present, most garden greening management still relies mainly on manual inspection and maintenance, which is not only inefficient, but also constrained by human resources and management costs. Traditional management methods cannot provide comprehensive and accurate garden data, resulting in the maintenance and management of green spaces often relying on empirical judgment and rough estimates. Therefore, in the technical solution of the present application, high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors are obtained, and combined with deep learning technology, a visual analysis report on the health status of the garden landscape is generated to fully understand the health status and potential problems of the garden, thereby implementing effective management and maintenance strategies.

[0028] Specifically, in the technical solution of the present application, the high-resolution green space coverage images taken by drones provide a detailed view of the green space, including vegetation coverage, health status, and potential disease problems. Drones can obtain large-scale, high-resolution images from the air, allowing garden managers to monitor changes in green spaces in real time. The garden landscape environment data collected by sensors include environmental parameters such as temperature, humidity, and soil quality. These data provide an additional dimension for garden management, allowing managers to understand the growth environment of green spaces. By integrating this information, garden managers can obtain more accurate green space status assessments, optimize maintenance and management strategies, reduce operating costs, make garden management more scientific and efficient, and help improve the quality and sustainability of urban green spaces.

[0029] In the above-mentioned digital garden landscape management system 100 based on visualization technology, the digital garden landscape data extraction module 120 is used to extract high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor. It should be understood that a comprehensive view of the garden green space and environmental conditions can be obtained by comprehensively analyzing these two feature vectors. Among them, the high-resolution image provides visual information, and the environmental data supplements the detailed data on the growth conditions. Combining the two can not only more accurately assess the health status of the garden, but also provide a scientific basis for garden maintenance, help optimize management decisions, reduce human errors, and improve the scientificity and accuracy of garden management. This comprehensive analysis method can better support garden health assessment, resource management and future planning, making garden management more intelligent and efficient.

[0030] Figure 2 FIG. 1 is a block diagram of a digital garden landscape data extraction module in a digital garden landscape management system based on visualization technology according to an embodiment of the present application. Figure 2 As shown, in a specific embodiment of the present application, the digital garden landscape data extraction module 120 includes: a green space coverage image feature extraction unit 121, used to perform feature extraction on the high-resolution green space coverage image taken by the drone to obtain the high-resolution green space coverage global correlation feature vector; a garden landscape environment data feature extraction unit 122, used to perform feature extraction on the garden landscape environment data collected by the sensor to obtain the garden landscape environment global correlation feature vector.

[0031] It should be understood that raw image data usually contains a lot of pixel information and noise, which makes it difficult to extract useful information directly from the image. Therefore, it is necessary to convert the image data into a more useful form, namely, a feature vector, through feature extraction. The main purpose of this process is to extract and encode key features in the image so that the important information in the image can be analyzed and interpreted more effectively. Among them, feature extraction involves identifying representative features from the image, such as vegetation type, distribution density, health status and its spatial distribution relationship. Through advanced image processing techniques and algorithms, these features are encoded into high-resolution green space coverage global correlation feature vectors. These feature vectors can simplify the raw image data and present complex visual information in a structured form, making further analysis more efficient and reliable. The extracted global correlation feature vector not only focuses on local features, but also takes into account all relevant spatial and feature information in the image. In this way, managers can get a holistic view of the entire green space, including vegetation health status, coverage and potential problem areas in different areas.

[0032] Furthermore, the garden environment data collected by sensors, including soil moisture, temperature, light intensity, and air quality, provide detailed information about environmental conditions. However, the raw sensor data usually contains a lot of noise and redundant information, and directly analyzing this data may lead to complexity and efficiency problems. Therefore, converting this data into a global correlation feature vector through feature extraction is an effective way to improve the efficiency of data processing and analysis. Among them, the main purpose of feature extraction is to extract the most representative and useful information from a large environmental data set. The feature vector can comprehensively consider the relationship between various environmental factors and capture the global patterns and trends in the data.

[0033] Figure 3 FIG. 1 is a block diagram of a green space coverage image feature extraction unit in a digital garden landscape management system based on visualization technology according to an embodiment of the present application. Figure 3 As shown, in a specific embodiment of the present application, the green space coverage image feature extraction unit 121 includes: a green space coverage data preprocessing subunit 1211, used to perform data preprocessing on the high-resolution green space coverage image taken by the drone to obtain a multi-region high-resolution green space coverage image block embedding vector; a green space coverage feature encoding subunit 1212, used to pass the multi-region high-resolution green space coverage image block embedding vector through a high-resolution green space coverage image block converter model based on the ViT model to obtain the high-resolution green space coverage global correlation feature vector.

[0034] It should be understood that the first step of data preprocessing is to divide the high-resolution image into multiple regional image blocks. This division allows for more detailed and localized processing, and each image block can be analyzed separately. Since drone images usually contain extensive ground information, directly processing the entire image may result in excessive computational burden, and details may be overlooked in some areas. By dividing the image into multiple small blocks, we can better focus on the characteristics of each area and improve the flexibility and efficiency of processing. Then, each image block is converted into a high-dimensional vector through embedding technology, which mathematically represents the feature information of the image block. This feature vector can capture key information in the image block, such as the type of vegetation, coverage density, health status, etc. This conversion simplifies complex visual data into an easy-to-process vector form.

[0035] Furthermore, the image patch embedding vectors of multi-region high-resolution green space coverage are passed through a high-resolution green space coverage image patch transformer model based on the ViT (Vision Transformer) model to achieve an advanced method of deep image understanding and data analysis. This process involves integrating local image information into global features to provide more comprehensive and accurate garden monitoring results. It should be understood that the image patch embedding vector provides local feature information extracted from the high-resolution green space coverage images taken by drones. Each image patch is converted into a feature vector, which contains details such as local vegetation coverage and health status. However, these local feature vectors cannot directly provide global information of the entire green space. Therefore, further processing is required to integrate these local features to obtain a comprehensive understanding of the entire green space. The image patch transformer model based on the ViT model plays a key role in this process. The ViT model is a visual model based on the Transformer architecture, which is good at processing global dependencies of image data. The ViT model regards the image patch embedding vector as sequence data and effectively captures the relationship between each block in the image through the self-attention mechanism. Among them, the self-attention mechanism can weight the features of each image block, thereby comprehensively considering local features and global contextual information. This enables the model to identify the correlation and importance between different regions, thereby generating a feature vector that can reflect global information. Through this processing method, a high-resolution green space coverage global correlation feature vector is formed. This global feature vector contains comprehensive information of the entire green space, representing the overall state of vegetation coverage, health status and its spatial distribution pattern. Compared with local features, the global feature vector provides a macro view of the entire green space condition, which can support more accurate analysis and decision-making.

[0036] In a specific embodiment of the present application, the green space coverage data preprocessing subunit 1211 includes: performing high-resolution green space coverage multi-region image segmentation on the high-resolution green space coverage image taken by the drone to obtain a high-resolution green space coverage multi-region image block sequence; and linearly embedding the high-resolution green space coverage multi-region image block sequence through a high-resolution green space coverage image block sequence encoder to obtain the multi-region high-resolution green space coverage image block embedding vector.

[0037] It should be understood that the high-resolution images taken by drones make each image very rich in details, but it also leads to the complexity of data processing. Directly analyzing the entire image may become computationally impractical due to the large amount of data, and some details may be overlooked. To solve this problem, it is an effective method to process the high-resolution image in blocks. The process of image blocking divides the entire high-resolution image into multiple smaller regional image blocks. This division not only reduces the size of each image block, but also makes the details of each block more concentrated, thereby reducing the complexity in processing. Each image block contains local information of the image, such as the type, density, and health status of vegetation in a specific area. The resulting multi-region image block sequence consists of multiple image blocks, each of which is a small part of the high-resolution image. By taking the image block sequence as input, a more detailed analysis can be performed.

[0038] Furthermore, the image patch sequence contains multiple local image patches segmented from high-resolution images taken by drones. Each image patch represents a part of the image and retains detailed information of the area. These local patches provide the basis for further analysis, but the original image patches are usually high-dimensional and difficult to use directly for model training or analysis. Therefore, it is necessary to convert these image patches into more operational feature representations. The linear embedding encoder is a tool that plays a role in this process. The main function of the linear embedding encoder is to map the high-dimensional image patch data to a low-dimensional embedding space. Through this mapping, each image patch is converted into a fixed-length vector that retains the key information in the image patch and has good operability and analyzability in the low-dimensional space. This mapping makes the data representation of the image patch more compact and can be used in a more efficient computing framework.

[0039] The image block embedding vectors processed by the linear embedding encoder of the high-resolution green space cover image block sequence effectively retain the characteristics of each image block while simplifying the data representation. These embedding vectors capture the main information in the image block, such as vegetation cover type, health status, and regional characteristics, thereby providing a comprehensive view of the image block and supporting more accurate assessment and management of the entire green space area.

[0040] Figure 4FIG. 1 is a block diagram of a garden landscape environment data feature extraction unit in a digital garden landscape management system based on visualization technology according to an embodiment of the present application. Figure 4 As shown, in a specific embodiment of the present application, the garden landscape environment data feature extraction unit 122 includes: a landscape environment data text understanding subunit 1221, which is used to pass the garden landscape environment data collected by the sensor through a garden landscape environment text understanding model to obtain multiple garden landscape environment data items; a landscape environment data feature encoding subunit 1222, which is used to feature encode the multiple garden landscape environment data items to obtain the garden landscape environment global correlation feature vector.

[0041] It should be understood that the garden landscape environment data collected by sensors is processed by the garden landscape environment text understanding model. Sensors are usually used in gardens to collect various environmental data, such as temperature, humidity, soil moisture, light intensity, etc. These data are original and huge, usually in unstructured or semi-structured form, and are difficult to use directly for analysis or decision-making. In order to make these data more usable, they must be converted into a more operational and explanatory form. Among them, the garden landscape environment text understanding model plays a key role in this process. Based on natural language processing technology, the sensor data is converted into structured environmental data items. Specifically, the model first preprocesses the data collected by the sensor to meet the needs of text analysis. Then, the model's text understanding ability is used to convert the data into specific environmental data items. Through this processing method, the original sensor data is effectively converted into more meaningful environmental data items. These data items not only contain the basic information of the original data, but also extract potential problems or trends in the data through the model's explanatory ability.

[0042] Furthermore, feature encoding is performed on multiple garden landscape environment data items to extract global information from the data and help make more comprehensive analysis and decision-making. Among them, in garden landscape management, the data items collected by sensors may include soil moisture, light intensity, air temperature, humidity, etc. These data items are usually independent and local feature information, and each data item provides a part of the environmental conditions. However, analyzing these data items individually may not provide a comprehensive understanding of the overall condition of the entire garden environment. Therefore, it is necessary to integrate these scattered data items through feature encoding. The process of feature encoding is to convert each data item into a more representative vector, which will capture the key information in the data item, thereby compressing the complex data items into a concise vector representation, so that these vectors can effectively reflect the relationship between the data items and the global characteristics of the environment.

[0043] In a specific embodiment of the present application, the landscape environment data feature encoding subunit 1222 includes: arranging the multiple garden landscape environment data items in two dimensions into a garden landscape environment information associated feature matrix; passing the garden landscape environment information associated feature matrix through a garden landscape environment information associated convolutional neural network as a feature encoder to obtain the garden landscape environment global associated feature vector.

[0044] It should be understood that in landscape management, multiple landscape environment data items may be related in time and space, and processing each data item separately may not reveal these complex relationships. In order to fully understand the state of the landscape environment, these data items need to be integrated into a systematic representation. Among them, the two-dimensional arrangement process is to arrange these multiple data items into a matrix in a certain order. This matrix is ​​usually displayed in the form of rows and columns, where rows can represent different time points or different spatial areas, and columns represent the characteristics of each data item. This arrangement can intuitively present the associated information of each data item at the same time point or area. For example, each row in the matrix can represent all environmental data at a specific time point, and each column represents the value of a specific environmental indicator at different time points. Arranging the data items into a matrix makes the data more structured and easy to process and analyze. The matrix form representation can clearly show the dimensions and relationships of the data, thereby simplifying data management.

[0045] Furthermore, in landscape management, the landscape environment information association feature matrix contains information extracted from multiple environmental data items (such as soil moisture, light intensity, etc.). The two-dimensional arrangement structure of the matrix makes the data easier to understand and analyze. However, this structured data may contain complex spatial and temporal dependencies, which are crucial to understanding the overall condition of the landscape environment. Landscape environment information association convolutional neural network (CNN) is used as a feature encoder to process these matrix data. CNN is a deep learning model that excels at extracting spatial features from images or similar structured data. Its core advantage is that it can automatically learn spatial and local patterns in data and convert these patterns into high-level feature representations. In this case, CNN is able to identify important patterns and associations in the feature matrix, such as how different environmental factors affect each other spatially or temporally. Through the convolution operation, CNN is able to apply a series of convolution kernels (filters) on the matrix, which can capture the local characteristics of the data. For example, when processing garden environment data, the convolution operation can reveal how soil moisture and light intensity jointly affect vegetation health. The features extracted by the convolutional layer are processed by the activation function and further passed to the pooling layer to reduce the dimension and retain important information.

[0046] Specifically, each layer of the garden landscape environment information associated convolutional neural network used as a feature encoder performs the following on the input data in the forward transfer of the layer: based on the convolution kernel, convolution processing is performed on the input data to generate a convolution feature map; global mean pooling processing is performed on the convolution feature map based on the feature matrix to generate a pooled feature map; and nonlinear activation is performed on the feature values ​​of each position in the pooled feature map to generate an activated feature addition map; wherein the output of the last layer of the garden landscape environment information associated convolutional neural network used as a feature encoder is the global associated feature vector of the garden landscape environment, the input from the second layer to the last layer of the garden landscape environment information associated convolutional neural network used as a feature encoder is the output of the previous layer, and the input of the garden landscape environment information associated convolutional neural network used as a feature encoder is the garden landscape environment information associated feature matrix.

[0047] In the above-mentioned digital garden landscape management system 100 based on visualization technology, the landscape health status report generation module 130 is used to generate a garden landscape health status visualization analysis report based on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector. It should be understood that the high-resolution green space coverage global correlation feature vector captures information such as green space distribution and plant coverage in the garden area, while the garden landscape environment global correlation feature vector summarizes the comprehensive characteristics of environmental factors such as soil moisture and light intensity. The two together reflect the green space health status and environmental conditions of the garden, providing rich data support for garden management. The process of generating a garden landscape health status visualization analysis report converts these complex global features into intuitive graphics and charts, so that garden managers can easily understand the health status of the garden. Through visualization technology, the report can display information such as the distribution of green space coverage and the changing trend of environmental conditions in the form of maps, heat maps, bar charts, etc. These visualization results can intuitively display the green space coverage, plant health status and the impact of environmental factors on green space in different areas of the garden. For example, a report could show the density of green space coverage in certain areas through a heat map, thus helping to identify areas with sparse green space.

[0048] Figure 5 FIG. 1 is a block diagram of a landscape health status report generation module in a digital garden landscape management system based on visualization technology according to an embodiment of the present application. Figure 5As shown, in a specific embodiment of the present application, the landscape health status report generation module 130 includes: a garden landscape feature fusion unit 131, which is used to perform feature fine-grained dimension offset adjustment on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector based on back-propagation information to obtain a garden landscape health status feature vector; a garden landscape health status visualization analysis report generation unit 132, which is used to pass the garden landscape health status feature vector through a generator to generate a garden landscape health status visualization analysis report.

[0049] It should be understood that the fusion of this high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector can comprehensively consider the influence of green space coverage and environmental factors to form a garden landscape health status feature vector. This fusion provides a comprehensive perspective to evaluate the health status of the garden by integrating the multi-dimensional information of green space coverage and environmental factors. The process of fusing feature vectors usually involves weighting or splicing the two global feature vectors so that the contribution of each vector can be effectively integrated. In this way, the obtained garden landscape health status feature vector can simultaneously reflect the comprehensive impact of green space coverage and environmental factors on garden health.

[0050] In particular, in the technical solution of the present application, the high-resolution green space coverage image generates a high-resolution green space coverage global correlation feature vector through image segmentation and linear embedding encoder, which captures the details and local features in the image. These features are further extracted through the ViT (Vision Transformer) model to extract global correlation information, reflecting the detailed structure, spatial layout and specific conditions of green space coverage, and emphasizing the detailed information of each block in the image and their positional relationship in the overall image. Relatively speaking, the garden landscape environment data is converted into data items through the text understanding model and processed by the convolutional neural network. The obtained garden landscape environment global correlation feature vector mainly focuses on the macro description and overall information of the environmental data. These features represent the comprehensive characteristics of the environment, such as climate conditions, soil types, etc., and the description of specific green space details is less, and more reflects the overall condition of the environment. Due to the significant differences in the dimensions and granularity of information processing between the high-resolution green space coverage feature vector and the garden landscape environment feature vector, direct simple fusion may lead to fine-grained dimension offset. Specifically, the high-resolution image feature vector has a higher spatial resolution and detail level, while the environmental feature vector contains more global information. This mismatch in information granularity may make it impossible to effectively align the details and global information of the two feature vectors during the fusion process, resulting in inconsistency in feature dimensions. Therefore, in the technical solution of the present application, the fine-grained dimension offset adjustment of the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector based on back-propagation information is performed to obtain the garden landscape health status feature vector.

[0051] Among them, the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector are subjected to feature fine-grained dimension offset adjustment based on back-propagation information to obtain the garden landscape health status feature vector, including: calculating the position difference between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a dimension offset differential feature vector; calculating the position point multiplication between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a dimension offset position point multiplication feature vector; calculating the position addition between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a dimension offset position addition feature vector; adding the dimension offset differential feature vector , the dimensional offset positional dot product feature vector and the dimensional offset positional addition feature vector are cascaded and input into a one-dimensional convolution layer and maximum pooling is performed to obtain a dimensional offset convolutional coding pooling feature vector; the square root of the two norm of the dimensional offset convolutional coding pooling feature vector is weighted and added with the one norm of the dimensional offset convolutional coding pooling feature vector to obtain a dimensional offset factor; the dimensional offset positional dot product feature vector and the square root of the length of the high-resolution green space coverage global correlation feature vector are added by position to obtain a dimensional offset intermediate vector; the dimensional offset factor is used as the first weighting coefficient and the second weighted hyperparameter is used as the second weighting coefficient to calculate the positional weighted sum of the dimensional offset positional addition feature vector and the dimensional offset intermediate vector to obtain a garden landscape health status feature vector.

[0052] Among them, the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector are subjected to feature fine-grained dimension offset adjustment based on back-propagation information to obtain the garden landscape health status feature vector, including:

[0053]

[0054] Among them, V1 represents the global correlation feature vector of high-resolution green space coverage, V2 represents the global correlation feature vector of garden landscape environment, represents positional subtraction, ⊙ represents positional multiplication, represents positional addition, conv1D represents a one-dimensional convolutional layer, MaxPool represents maximum pooling, V μ represents the dimension-shifted convolutional coding pooling feature vector, ||·||1 represents the first norm, ||·||2 represents the second norm, [·] represents the cascade, α represents the first weighted hyperparameter, τ represents the dimension-shift factor, L represents the length of the high-resolution green space coverage global correlation feature vector, β represents the second weighted hyperparameter, V f Represents the characteristic vector of the health status of the garden landscape.

[0055] In the technical solution of the present application, in the process of fusing the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector, the feature fine-grained dimension offset between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector is taken into account, which may cause the fused garden landscape health status feature vector to have local structural collapse. Based on this, in the technical solution of the present application, the feature fine-grained dimension offset adjustment based on the back-propagation information is performed on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector.

[0056] Specifically, based on the restricted expression of the foreground-background structure distinction of the high-dimensional manifold of the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector by the amplitude of the semantic difference vector between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector, the fine-grained correspondence of key features between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector is modeled, and then active fine-grained correspondence adjustment propagation is performed through the back-propagation displacement compensation information between the corresponding eigenvalues ​​of the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector, so as to avoid the imbalance at the feature vector level of the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector in a focusing manner, thereby synchronizing the micro dimensions of the two feature vectors, ensuring that during the fusion process, the information between the feature vectors can be effectively aligned and combined, thereby avoiding the abnormal accumulation of local structures.

[0057] Furthermore, the role of the generator in this process is to convert these feature vector data into visual analysis reports. Through visual display, managers can make more accurate management decisions based on intuitive data and trends, such as adjusting irrigation strategies or increasing vegetation cover. Among them, the generator may refer to a generative adversarial network (GAN) or an autoencoder, which can convert the feature vectors of the garden health status into visual graphics or reports. Specifically, the generative adversarial network (GAN) generates realistic images and reports through adversarial training, and the autoencoder generates relevant visualizations through data reconstruction technology. These generators convert complex feature data into easy-to-understand and easy-to-analyze graphical displays, helping users to intuitively assess the health of the garden landscape.

[0058] In summary, the embodiment of the present application first obtains high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors, and then uses deep learning technology to perform feature extraction and correlation analysis on the two. Finally, a generator is used to generate a visual analysis report on the health status of the garden landscape to understand the health status and potential risks of the garden, and then take effective management and maintenance measures.

[0059] As described above, the digital garden landscape management system 100 based on visualization technology according to the embodiment of the present application can be implemented in various terminal devices. In one example, the digital garden landscape management system 100 based on visualization technology can be integrated into the terminal device as a software module and / or a hardware module. For example, the digital garden landscape management system 100 based on visualization technology can be a software module in the operating system of the terminal device, or can be an application developed for the terminal device; of course, the digital garden landscape management system 100 based on visualization technology can also be one of the many hardware modules of the terminal device.

[0060] Alternatively, in another example, the digital garden landscape management system 100 based on visualization technology and the terminal device may also be separate devices, and the digital garden landscape management system 100 based on visualization technology may be connected to the terminal device via a wired and / or wireless network, and transmit interactive information in accordance with an agreed data format.

[0061] Figure 6 FIG. 1 is a flow chart of a digital garden landscape management method based on visualization technology according to an embodiment of the present application. Figure 6 As shown, according to the digital garden landscape management method based on visualization technology according to the embodiment of the present application, it includes: S110, obtaining a high-resolution green space coverage image taken by a drone and garden landscape environment data collected by a sensor; S120, extracting a high-resolution green space coverage global correlation feature vector and a garden landscape environment global correlation feature vector from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor; S130, generating a garden landscape health status visualization analysis report based on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector.

[0062] Here, those skilled in the art can understand that the specific operations of each step in the above-mentioned digital garden landscape management method based on visualization technology have been referred to above. Figures 1 to 5 The description of the digital garden landscape management system based on visualization technology has been introduced in detail, and therefore, its repeated description will be omitted.

[0063] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0064] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0065] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0066] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0067] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0068] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The second and other words are used to indicate names, but not to indicate any particular order.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit of the technical solution of the present invention.

Claims

1. A digital garden landscape management system based on visualization technology, characterized in that: include: Digital garden landscape data acquisition module, used to obtain high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors; A digital garden landscape data extraction module is used to extract high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor; The landscape health status report generation module is used to generate a visual analysis report on the garden landscape health status based on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector.

2. The digital garden landscape management system based on visualization technology according to claim 1 is characterized in that: The digital garden landscape data extraction module comprises: A green space coverage image feature extraction unit, used for performing feature extraction on the high-resolution green space coverage image taken by the drone to obtain the high-resolution green space coverage global correlation feature vector; The garden landscape environment data feature extraction unit is used to extract features from the garden landscape environment data collected by the sensor to obtain a global correlation feature vector of the garden landscape environment.

3. The digital garden landscape management system based on visualization technology according to claim 2 is characterized in that: The green space coverage image feature extraction unit comprises: A green space coverage data preprocessing subunit is used to perform data preprocessing on the high-resolution green space coverage image taken by the drone to obtain multi-region high-resolution green space coverage image block embedding vectors; The green space coverage feature encoding subunit is used to embed the multi-region high-resolution green space coverage image block into a vector through a high-resolution green space coverage image block converter model based on the ViT model to obtain the high-resolution green space coverage global correlation feature vector.

4. The digital garden landscape management system based on visualization technology according to claim 3 is characterized in that: The green space coverage data preprocessing subunit includes: Performing high-resolution green space coverage multi-region image block division on the high-resolution green space coverage image taken by the drone to obtain a high-resolution green space coverage multi-region image block sequence; The high-resolution green space coverage multi-region image block sequence is linearly embedded into an encoder through a high-resolution green space coverage image block sequence to obtain an embedding vector of the multi-region high-resolution green space coverage image block.

5. The digital garden landscape management system based on visualization technology according to claim 4 is characterized in that: The garden landscape environment data feature extraction unit comprises: A landscape environment data text understanding subunit, used for passing the garden landscape environment data collected by the sensor through a garden landscape environment text understanding model to obtain a plurality of garden landscape environment data items; The landscape environment data feature encoding subunit is used to perform feature encoding on the multiple garden landscape environment data items to obtain the garden landscape environment global correlation feature vector.

6. The digital garden landscape management system based on visualization technology according to claim 5 is characterized in that: The landscape environment data feature encoding subunit includes: Arranging the plurality of garden landscape environment data items in two dimensions into a garden landscape environment information associated feature matrix; The garden landscape environment information association feature matrix is ​​passed through a garden landscape environment information association convolutional neural network as a feature encoder to obtain the garden landscape environment global association feature vector.

7. The digital garden landscape management system based on visualization technology according to claim 6 is characterized in that: The landscape health status report generation module includes: A garden landscape feature fusion unit is used to perform feature fine-grained dimension offset adjustment based on back-propagation information on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a garden landscape health status feature vector; The garden landscape health status visualization analysis report generating unit is used to pass the garden landscape health status feature vector through a generator to generate a garden landscape health status visualization analysis report.

8. The digital garden landscape management system based on visualization technology according to claim 7 is characterized in that: The garden landscape feature fusion unit comprises: Calculating the positional difference between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a dimensionally offset differential feature vector; Calculate the position point multiplication between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a dimension offset position point multiplication feature vector; Calculate the positional addition between the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector to obtain a dimension offset positional addition feature vector; The dimension offset differential feature vector, the dimension offset position point product feature vector and the dimension offset position addition feature vector are cascaded and input into a one-dimensional convolution layer and maximum pooling is performed to obtain a dimension offset convolutional coding pooling feature vector; The square root value of the two-norm of the dimension-shifted convolutional coding pooling feature vector is weighted and added to the one-norm of the dimension-shifted convolutional coding pooling feature vector to obtain a dimension-shift factor; Adding the positional dot product feature vector of the dimension offset and the square root of the length of the high-resolution green space coverage global correlation feature vector according to the position to obtain a dimension offset intermediate vector; The dimension shift factor is used as the first weighting coefficient, and the second weighted hyperparameter is used as the second weighting coefficient to calculate the position-weighted sum of the dimension shift position-added feature vector and the dimension shift intermediate vector to obtain the garden landscape health status feature vector.

9. A digital garden landscape management method based on visualization technology, characterized in that: include: Obtain high-resolution green space coverage images taken by drones and garden landscape environment data collected by sensors; Extracting high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor; Based on the high-resolution green space coverage global correlation feature vector and the garden landscape environment global correlation feature vector, a garden landscape health status visualization analysis report is generated.

10. The digital garden landscape management method based on visualization technology according to claim 9 is characterized in that: Extracting high-resolution green space coverage global correlation feature vectors and garden landscape environment global correlation feature vectors from the high-resolution green space coverage image taken by the drone and the garden landscape environment data collected by the sensor, including: Performing feature extraction on the high-resolution green space coverage image taken by the drone to obtain a global correlation feature vector of the high-resolution green space coverage; Feature extraction is performed on the garden landscape environment data collected by the sensor to obtain a global correlation feature vector of the garden landscape environment.

Images