A method and system for generating the eight scenic spots of a city based on deep learning technology

Abstract

This invention discloses a method and system for generating eight urban scenic spots based on deep learning technology, which relates to the field of landscape architecture. The method includes collecting excellent photographs of urban landscapes and screening out landscape images belonging to the traditional eight scenic spots using a neural network model. The neural network model then scores the screened landscape images based on predefined scoring criteria to determine the top eight landscape images. The neural network model then names the landscapes in the identified landscape images based on predefined naming rules to generate the new eight urban scenic spots. The invention can automatically determine and name the eight urban scenic spots efficiently and accurately.

Description

Technical Field

[0001] The present invention relates to the field of landscape architecture, and in particular to a method and system for generating eight urban scenes based on deep learning technology. Background Art

[0002] Machine learning is a key branch of artificial intelligence. It allows computers to automatically learn from human experience through data, and deep learning is a key branch of this field. Deep learning is a multi-layered neural network structure inspired by the neural network structure of the human brain. Deep learning can achieve accurate fit to data distributions through training. Compared to other machine learning methods, deep learning does not require complex feature engineering and can directly learn the inherent patterns of data through studying training data, making it more intelligent.

[0003] The Eight Scenic Spots of a City is a cultural phenomenon, a collective term for eight scenic spots commonly recognized in a region, encompassing both natural and cultural landscapes. Currently, the selection and evaluation of these eight scenic spots is typically done manually, which is time-consuming, labor-intensive, and inefficient. Summary of the Invention

[0004] In view of the defects in the prior art, the purpose of the present invention is to provide a method and system for generating eight urban scenes based on deep learning technology, which can realize the automatic determination and naming of the eight urban scenes in an efficient and accurate manner.

[0005] To achieve the above objectives, the present invention provides a method for generating eight urban scenes based on deep learning technology, which specifically includes the following steps:

[0006] Collect excellent photographs of urban landscapes and select the landscape images that belong to the traditional eight scenic spots based on the neural network model;

[0007] Based on the set scoring criteria, the selected landscape images are scored using a neural network model to determine the top eight landscape images;

[0008] Based on the set naming rules, the landscapes in the determined landscape pictures are named through the neural network model to obtain the new eight scenic spots of the city.

[0009] On the basis of the above technical solution, the steps of selecting landscape pictures belonging to the traditional eight scenic spots based on the neural network model include:

[0010] Obtain pictures of the traditional eight scenic spots as a training set, train the neural network model, and obtain an artificial intelligence eight scenic spot screening network model;

[0011] Through the artificial intelligence eight-scene screening network model, the collected excellent photographs of urban landscapes are screened to obtain landscape pictures belonging to the traditional eight scenes.

[0012] On the basis of the above technical solution, the selected landscape pictures are scored by a neural network model based on the set scoring criteria to determine the top eight landscape pictures. The specific steps include:

[0013] Collect online landscape pictures and actively rate them based on aesthetics, interests, and preferences;

[0014] The scored online landscape images are used as a training set to train the neural network model, and an artificial intelligence image sentiment evaluation network model is obtained;

[0015] The selected landscape images are scored using an artificial intelligence image sentiment evaluation network model, and landscape images with scores greater than a preset score are retained;

[0016] The retained landscape pictures were ranked according to their frequency of appearance, web page comments, and number of likes, and the top eight landscape pictures were determined.

[0017] On the basis of the above technical solutions,

[0018] The types of the traditional eight scenes include object type and landscape type;

[0019] The object classes include pavilions, terraces, towers, shepherd boys, bridges, animals, fishermen and fishermen's songs;

[0020] The landscape categories include scenery of four seasons, time, weather, ancient trees, temples, mountain colors, flowers and plants, water colors, ancient ferries, streets, hot springs, pools and mountain villages.

[0021] On the basis of the above technical solution, the method of naming the landscapes in the determined landscape pictures based on the set naming rules by using a neural network model to obtain the new eight city scenes includes the following specific steps:

[0022] Obtain existing traditional eight scenic spots pictures, classify them, and label the pictures by type;

[0023] Input the type-labeled images into the neural network model to train the neural network model;

[0024] Match the types of the traditional eight scenic spots with the corresponding ancient landscape phrases to obtain a knowledge graph that can intelligently describe the landscape according to the landscape type;

[0025] According to the place names of the landscapes in the landscape pictures and the description of the landscapes in the landscape pictures by the knowledge graph, the eight titles of the landscapes in the landscape pictures are determined to obtain the new eight scenic spots of the city.

[0026] Based on the above technical solution, the location name of the landscape in the landscape image is determined as follows:

[0027] When the name of the landscape location in the landscape image is two characters, the name of the landscape location is used as the location name of the landscape;

[0028] When the name of the location of the landscape in the landscape image is more than two characters, two characters in the name of the location of the landscape are selected as the location name of the landscape;

[0029] When the name of the landscape location in the landscape image is one word, the name of the landscape location is combined with the type name obtained by the intelligent landscape description neural network model to classify the landscape as the location name of the landscape.

[0030] The present invention provides a system for generating eight urban scenes based on deep learning technology, comprising:

[0031] Picture collection module, which is used to collect excellent photographic works about urban landscapes;

[0032] A picture screening module is used to screen the excellent photographic works on urban landscapes collected by the picture collection module and select landscape pictures belonging to the traditional eight scenic spots;

[0033] The picture rating module is used to rate the landscape pictures selected by the picture screening module based on the set rating criteria and determine the top eight landscape pictures;

[0034] The eight-scene naming module is used to name the landscapes in the landscape pictures determined by the picture rating module based on the set naming rules to obtain the new eight scenic spots of the city.

[0035] Based on the above technical solution, the picture screening module screens out landscape pictures belonging to the traditional eight scenic spots. The specific process includes:

[0036] Obtain pictures of the traditional eight scenic spots as a training set, train the neural network model, and obtain an artificial intelligence eight scenic spot screening network model;

[0037] Through the artificial intelligence eight-scene screening network model, the collected excellent photographs of urban landscapes are screened to obtain landscape pictures belonging to the traditional eight scenes.

[0038] Based on the above technical solution, the picture rating module scores the landscape pictures selected by the picture screening module and determines the top eight landscape pictures. The specific process includes:

[0039] Collect online landscape pictures and actively rate them based on aesthetics, interests, and preferences;

[0040] The scored online landscape images are used as a training set to train the neural network model, and an artificial intelligence image sentiment evaluation network model is obtained;

[0041] The selected landscape images are scored using an artificial intelligence image sentiment evaluation network model, and landscape images with scores greater than a preset score are retained;

[0042] The retained landscape pictures were ranked according to their frequency of appearance, web page comments, and number of likes, and the top eight landscape pictures were determined.

[0043] On the basis of the above technical solution, the eight sceneries naming module names the sceneries in the sceneries pictures determined by the picture rating module based on the set naming rules to obtain the new eight sceneries of the city. The specific process includes:

[0044] Obtain existing traditional eight scenic spots pictures, classify them, and label the pictures by type;

[0045] Input the type-labeled images into the neural network model to train the neural network model;

[0046] Match the types of the traditional eight scenic spots with the corresponding ancient landscape phrases to obtain a knowledge graph that can intelligently describe the landscape according to the landscape type;

[0047] According to the place names of the landscapes in the landscape pictures and the description of the landscapes in the landscape pictures by the knowledge graph, the eight titles of the landscapes in the landscape pictures are determined to obtain the new eight scenic spots of the city.

[0048] Compared with the existing technology, the advantages of the present invention are: by collecting excellent photographic works about urban landscapes, and screening out landscape pictures belonging to the traditional eight scenes based on a neural network model, and then scoring the screened landscape pictures based on set scoring criteria through the neural network model to determine the top eight landscape pictures, and then naming the landscapes in the determined landscape pictures based on set naming rules through the neural network model to obtain the new eight scenes of the city, thereby realizing automatic determination and naming of the eight scenes of the city, which is efficient and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0050] Figure 1This is a flowchart of a method for generating eight urban scenes based on deep learning technology in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments.

[0052] See also Figure 1 As shown, an embodiment of the present invention provides a method for generating eight scenic spots of a city based on deep learning technology. The method searches for landscape images of the city to be evaluated on the Internet and retains the address information of the landscape in the images. The artificial intelligence eight scenic spot screening network model is used to select landscape images belonging to the traditional eight scenic spots. The artificial intelligence picture emotion evaluation network model is used to retain high-quality eight scenic spot landscape images, thereby selecting eight scenic spots. The method for generating eight scenic spots of a city specifically includes the following steps:

[0053] S1: Collect excellent photographs of urban landscapes, and select landscape images belonging to the traditional eight scenes based on a neural network model; excellent photographs of urban landscapes can be collected on the Internet.

[0054] In an embodiment of the present invention, landscape pictures belonging to the traditional eight scenic spots are screened out based on a neural network model, and the specific steps include:

[0055] S101: Obtain pictures of the traditional eight scenic spots as a training set, train a neural network model, and obtain an artificial intelligence eight scenic spot screening network model; the neural network model here can be a ResNet neural network (residual neural network). By training the neural network model, the trained artificial intelligence eight scenic spot screening network model has the ability to identify the traditional eight scenic spots, thereby identifying whether the landscape in the picture belongs to the traditional eight scenic spots.

[0056] In the present invention, the types of the traditional eight scenes include object class and landscape class; the object class includes pavilions, platforms, towers, shepherd boys, bridges, animals, fishermen and fisherman songs; the landscape class includes four seasons, time, weather, ancient trees, temples, mountain colors, flowers and plants, water colors, ancient ferries, streets, hot springs, pools and mountain villages.

[0057] S102: Filter the collected excellent photographs of urban landscapes through the artificial intelligence eight-scene screening network model to obtain landscape pictures belonging to the traditional eight scenes.

[0058] S2: Based on the set scoring criteria, the selected landscape images are scored through the neural network model to determine the top eight landscape images;

[0059] In an embodiment of the present invention, based on a set scoring standard, the filtered landscape pictures are scored by a neural network model to determine the top eight landscape pictures. The specific steps include:

[0060] S201: Collect online landscape pictures, and actively score the collected online landscape pictures based on aesthetics, interests and preferences; that is, use a manual scoring method, with 10 points as the full score, to score the collected online landscape pictures.

[0061] S202: Using the scored online landscape pictures as a training set to train a neural network model, thereby obtaining an artificial intelligence picture emotion evaluation network model; by training the neural network model, the neural network model is enabled to have the ability to score the landscape in the picture.

[0062] S203: Scoring the selected landscape images using an artificial intelligence image sentiment evaluation network model, and retaining landscape images with scores greater than a preset score;

[0063] S204: Rank the retained landscape pictures according to their appearance frequency, web page comments, and number of likes, and determine the top eight landscape pictures.

[0064] S3: Based on the set naming rules, the neural network model is used to name the landscapes in the determined landscape image to obtain the new eight scenic spots of the city. In the present invention, the eight scenic spots of the city are named using an A+B model, where A is the location name of the landscape and B is a landscape description phrase, so that the eight scenic spots of the city are named as Baishui Xiaodu, Zishan Dongcui, Hanjiang Yalu, etc.

[0065] In an embodiment of the present invention, based on a set naming rule, a neural network model is used to name the landscapes in the determined landscape image to obtain the new eight city scenes. The specific steps include:

[0066] S301: Obtain existing pictures of the eight traditional scenic spots, classify them, and label the pictures by type; for example, picture a is a pavilion in the object class, and picture b is a mountain scene in the landscape class.

[0067] S302: Inputting the type-labeled image into a neural network model to train the neural network model;

[0068] S303: Match the types of the eight traditional scenic spots with corresponding ancient landscape phrases to generate a knowledge graph that can intelligently describe the landscape based on the landscape type. The trained neural network model identifies the landscape type, and the knowledge graph matches the landscape type with ancient landscape phrases based on the landscape type. Ancient landscape phrases include "spring rain," "autumn clear," "hidden snow," and "embracing green," among others. For example, if the AI landscape description network model identifies image b as a mountain scene, it can match "mountain scene" with "embracing green" from the ancient landscape phrase to describe the mountain scene. If the AI landscape description network model identifies image c as a temple scene, it can match "temple" with "bell sound" from the ancient landscape phrase to describe the temple scene.

[0069] S304: Based on the location name of the landscape in the landscape image and the description of the landscape in the landscape image in the knowledge graph, the eight scenic spots in the landscape image are determined to obtain the new eight scenic spots of the city. For example, if the type of landscape image b is mountain scenery, the location of the landscape in landscape image b is Longshan, and the descriptive phrase corresponding to the type of landscape image b is Yongcui, then the name of the landscape in landscape image b is Longshan Yongcui.

[0070] It should be noted that the location name of the landscape in the landscape image is determined as follows:

[0071] When the name of the landscape location in the landscape image is two characters, the name of the landscape location is used as the location name of the landscape;

[0072] When the name of the location of the landscape in the landscape image is more than two characters, two characters in the name of the location of the landscape are selected as the place name of the landscape; for example, if the name of the location of the landscape is "Wuhan Huangpi Mulan", "Mulan" is used as the place name of the landscape.

[0073] When the name of the location in a landscape image is a single character, the location name is combined with the type name obtained by the intelligent landscape description neural network model to form the location name of the landscape. For example, if the location of landscape image c is E and the type of landscape in landscape image c is mountain, then the location name of the landscape in landscape image c is Eshan.

[0074] The identification of the new eight scenic spots in this invention is primarily based on a deep learning neural network model. This neural network model can classify images and segment image content based on given conditions, and can use an image judgment mechanism similar to human emotional landscape aesthetics. The eight scenic spot data inherits the aesthetic tradition of ancient landscapes. Through big data, urban landscape images from online platforms are captured and preliminarily screened for aesthetics and landscape image types using artificial intelligence. From these screened images, the top eight most popular landscapes are selected based on their online frequency. Using an intelligent eight-scenic spot naming system, the new eight scenic spots are selected and named.

[0075] The method for generating eight urban scenes based on deep learning technology in an embodiment of the present invention collects excellent photographic works about urban landscapes, and screens out landscape pictures belonging to the traditional eight scenes based on a neural network model. Then, based on set scoring criteria, the screened landscape pictures are scored by the neural network model to determine the top eight landscape pictures. Then, based on set naming rules, the landscapes in the determined landscape pictures are named by the neural network model to obtain the new eight urban scenes, thereby realizing automatic determination and naming of the eight urban scenes in an efficient and accurate manner.

[0076] An embodiment of the present invention provides a system for generating eight urban scenic spots based on deep learning technology, which includes a picture collection module, a picture screening module, a picture rating module and an eight-scene naming module.

[0077] The picture collection module is used to collect excellent photographic works about urban landscapes; the picture screening module is used to screen the excellent photographic works about urban landscapes collected by the picture collection module, and screen out landscape pictures belonging to the traditional eight scenic spots; the picture rating module is used to score the landscape pictures screened by the picture screening module based on the set scoring criteria, and determine the top eight landscape pictures; the eight-scene naming module is used to name the landscapes in the landscape pictures determined by the picture rating module based on the set naming rules, and obtain the new eight scenic spots of the city.

[0078] In the present invention, the picture screening module screens out landscape pictures belonging to the traditional eight scenic spots. The specific process includes:

[0079] Obtain pictures of the traditional eight scenic spots as a training set, train the neural network model, and obtain an artificial intelligence eight scenic spot screening network model;

[0080] Through the artificial intelligence eight-scene screening network model, the collected excellent photographs of urban landscapes are screened to obtain landscape pictures belonging to the traditional eight scenes.

[0081] In the present invention, the picture rating module scores the landscape pictures screened by the picture screening module and determines the top eight landscape pictures. The specific process includes:

[0082] Collect online landscape pictures and actively rate them based on aesthetics, interests, and preferences;

[0083] The scored online landscape images are used as a training set to train the neural network model, and an artificial intelligence image sentiment evaluation network model is obtained;

[0084] The selected landscape images are scored using an artificial intelligence image sentiment evaluation network model, and landscape images with scores greater than a preset score are retained;

[0085] The retained landscape pictures were ranked according to their frequency of appearance, web page comments, and number of likes, and the top eight landscape pictures were determined.

[0086] In the present invention, the eight scene naming module names the sceneries in the scene pictures determined by the picture rating module based on the set naming rules to obtain the new eight sceneries of the city. The specific process includes:

[0087] Obtain existing traditional eight scenic spots pictures, classify them, and label the pictures by type;

[0088] Input the type-labeled images into the neural network model to train the neural network model;

[0089] Match the types of the traditional eight scenic spots with the corresponding ancient landscape phrases to obtain a knowledge graph that can intelligently describe the landscape according to the landscape type;

[0090] According to the place names of the landscapes in the landscape pictures and the description of the landscapes in the landscape pictures by the knowledge graph, the eight titles of the landscapes in the landscape pictures are determined to obtain the new eight scenic spots of the city.

[0091] The foregoing is merely a list of specific embodiments of the present application, intended to enable those skilled in the art to understand or implement the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the broadest scope consistent with the principles and novel features of the present application.

[0092] 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 flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, 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 flowcharts and / or block diagrams. 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.

Claims

1. A method for generating eight urban scenes based on deep learning technology, characterized in that: The specific steps include: Collect excellent photographs of urban landscapes and select the landscape images that belong to the traditional eight scenic spots based on the neural network model; Based on the set scoring criteria, the selected landscape images are scored using a neural network model to determine the top eight landscape images; Based on the set naming rules, the neural network model is used to name the landscapes in the determined landscape pictures to obtain the new eight scenic spots of the city; The steps of scoring the selected landscape images based on the set scoring criteria using a neural network model to determine the top eight landscape images include: Collect online landscape pictures and actively rate them based on aesthetics, interests, and preferences; The scored online landscape images are used as a training set to train the neural network model, and an artificial intelligence image sentiment evaluation network model is obtained; The selected landscape images are scored using an artificial intelligence image sentiment evaluation network model, and landscape images with scores greater than a preset score are retained; Rank the retained landscape images according to their frequency of appearance, web page comments, and number of likes, and determine the top eight landscape images; The steps of naming the landscapes in the determined landscape pictures based on the set naming rules by using a neural network model to obtain the new eight city scenes include: Obtain existing traditional eight scenic spots pictures, classify them, and label the pictures by type; Input the type-labeled images into the neural network model to train the neural network model; Match the types of the traditional eight scenic spots with the corresponding ancient landscape phrases to obtain a knowledge graph that can intelligently describe the landscape according to the landscape type; According to the place names of the landscapes in the landscape pictures and the descriptions of the landscapes in the landscape pictures in the knowledge graph, the eight scenic spots in the landscape pictures are determined to obtain the new eight scenic spots of the city; The location name of the landscape in the landscape image is determined as follows: When the name of the landscape location in the landscape image is two characters, the name of the landscape location is used as the location name of the landscape; When the name of the location of the landscape in the landscape image is more than two characters, two characters in the name of the location of the landscape are selected as the location name of the landscape; When the name of the landscape location in the landscape image is one word, the name of the landscape location is combined with the type name obtained by the intelligent landscape description neural network model to classify the landscape as the location name of the landscape.

2. The method for generating eight urban scenes based on deep learning technology according to claim 1, characterized in that: The steps of selecting landscape images belonging to the traditional eight scenic spots based on the neural network model include: Obtain pictures of the traditional eight scenic spots as a training set, train the neural network model, and obtain an artificial intelligence eight scenic spot screening network model; Through the artificial intelligence eight-scene screening network model, the collected excellent photographs of urban landscapes are screened to obtain landscape pictures belonging to the traditional eight scenes.

3. The method for generating eight urban scenes based on deep learning technology according to claim 1, characterized in that: The types of the traditional eight scenes include object type and landscape type; The object classes include pavilions, terraces, towers, shepherd boys, bridges, animals, fishermen and fishermen's songs; The landscape categories include scenery of four seasons, time, weather, ancient trees, temples, mountain colors, flowers and plants, water colors, ancient ferries, streets, hot springs, pools and mountain villages.

4. A system for generating eight urban scenes based on deep learning technology, characterized in that: include: Picture collection module, which is used to collect excellent photographic works about urban landscapes; A picture screening module is used to screen the excellent photographic works on urban landscapes collected by the picture collection module and select landscape pictures belonging to the traditional eight scenic spots; The picture rating module is used to rate the landscape pictures selected by the picture screening module based on the set rating criteria and determine the top eight landscape pictures; The eight-scene naming module is used to name the landscapes in the landscape pictures determined by the picture rating module based on the set naming rules to obtain the new eight scenic spots of the city; The picture rating module scores the landscape pictures selected by the picture screening module and determines the top eight landscape pictures. The specific process includes: Collect online landscape pictures and actively rate them based on aesthetics, interests, and preferences; The scored online landscape images are used as a training set to train the neural network model, and an artificial intelligence image sentiment evaluation network model is obtained; The selected landscape images are scored using an artificial intelligence image sentiment evaluation network model, and landscape images with scores greater than a preset score are retained; Rank the retained landscape images according to their frequency of appearance, web page comments, and number of likes, and determine the top eight landscape images; The eight-scene naming module names the landscapes in the landscape pictures determined by the picture rating module based on the set naming rules to obtain the new eight city scenes. The specific process includes: Obtain existing traditional eight scenic spots pictures, classify them, and label the pictures by type; Input the type-labeled images into the neural network model to train the neural network model; Match the types of the traditional eight scenic spots with the corresponding ancient landscape phrases to obtain a knowledge graph that can intelligently describe the landscape according to the landscape type; According to the place names of the landscapes in the landscape pictures and the descriptions of the landscapes in the landscape pictures in the knowledge graph, the eight scenic spots in the landscape pictures are determined to obtain the new eight scenic spots of the city; The location name of the landscape in the landscape image is determined as follows: When the name of the landscape location in the landscape image is two characters, the name of the landscape location is used as the location name of the landscape; When the name of the location of the landscape in the landscape image is more than two characters, two characters in the name of the location of the landscape are selected as the location name of the landscape; When the name of the landscape location in the landscape image is one word, the name of the landscape location is combined with the type name obtained by the intelligent landscape description neural network model to classify the landscape as the location name of the landscape.

5. The system for generating eight urban scenes based on deep learning technology according to claim 4, characterized in that: The picture screening module screens out landscape pictures belonging to the traditional eight scenic spots. The specific process includes: Obtain pictures of the traditional eight scenic spots as a training set, train the neural network model, and obtain an artificial intelligence eight scenic spot screening network model; Through the artificial intelligence eight-scene screening network model, the collected excellent photographs of urban landscapes are screened to obtain landscape pictures belonging to the traditional eight scenes.

Images