One-stop service platform based on artificial intelligence and application thereof

Abstract

The invention relates to the technical field of artificial intelligence, and discloses a one-stop service platform based on artificial intelligence and application thereof, which is a platform integrating intelligent naming, intelligent logo design, intelligent poster design, an intelligent matting technology and an intellectual property management system, applies deep learning, natural language processing and image recognition technologies, combines big data analysis, and provides an intellectual property management platform. And comprehensive brand creation and intellectual property management services are provided for small and medium-sized enterprises. According to the platform, functions of intelligent naming, design and the like are integrated, so that the brand creation and market response speed of small and medium-sized enterprises is remarkably improved; the automatic and intelligent service of the platform improves the market competitiveness of small and medium-sized enterprises, and reduces the threshold of professional technical support.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an artificial intelligence-based one-stop service platform and its application. Background Art

[0002] With the rapid development of big data and artificial intelligence technologies, small and medium-sized enterprises are increasingly demanding integrated intelligent tool application platforms. In the traditional design service model, the cooperation between demanders and designers often faces problems of poor communication and misunderstanding. Designers need to repeatedly interpret the intentions of demanders, and demanders often find it difficult to accurately express their design needs. This information asymmetry leads to frequent program modifications, which not only prolongs the design cycle and increases costs, but may also cause the final design results to be far from the original intention. In addition, the graphic registration risk assessment after the design is completed is also a complex and time-consuming process that requires professional knowledge and experience to evaluate the originality and registrability of the design, which is an insurmountable obstacle for most small and medium-sized enterprises.

[0003] Small and medium-sized enterprises usually lack professional technical support in naming, brand design, data analysis, etc. Naming requires not only creativity and market insight, but also multiple factors such as culture, law and business registration. Brand design needs to combine brand positioning, market trends and consumer preferences, which requires designers to have deep professional knowledge and innovation capabilities. Data analysis involves the collection, processing and interpretation of market data, which is crucial for formulating effective market strategies. However, these professional services are often expensive, and small and medium-sized enterprises find it difficult to obtain high-quality services due to budget and resource constraints. Summary of the invention

[0004] In order to solve the above problems, this patent proposes an artificial intelligence-based one-stop service platform and its application, aiming to improve service efficiency, reduce design costs, and provide trademark registration risk assessment.

[0005] The technical solution of the present invention is as follows:

[0006] An AI-based one-stop service platform that integrates at least:

[0007] Intelligent naming system: By using LSTM model and big data, combined with industry information and / or legal person information, online naming is realized;

[0008] Intelligent logo design system: Through NLP and sequence planning technology, it can quickly generate and edit logo solutions based on brand information and preferences, including generating VI scene displays and business card designs;

[0009] Intelligent poster design system: using the established material center and deep reinforcement learning to automatically design personalized posters;

[0010] Intelligent cutout system: uses convolutional neural network to predict the alpha channel of the image from the original image and the three-dimensional image to achieve accurate cutout of the image;

[0011] Intellectual Property Management System: Combined with AI, it provides comprehensive online intellectual property services, including management, monitoring and registration.

[0012] Furthermore, the intelligent naming system comprises:

[0013] A deep learning LSTM neural network model is used to link at least 90 million industrial and commercial enterprise information databases and at least 300,000 Tang and Song poetry libraries, and candidate company names are generated based on the company's industry information and / or company legal person information entered by the user.

[0014] Furthermore, the intelligent naming system also includes:

[0015] Based on the multi-dimensional analysis of the big data of registered enterprise logos and information of enterprises to be registered, as well as the comparison and duplication test of industrial and commercial registration, candidate enterprise names with batch and high industrial and commercial registration pass rates are generated.

[0016] Furthermore, the intelligent logo design system includes:

[0017] Using deep learning natural language processing technology and deep learning sequence planning network model technology, it generates multiple sets of alternative logo graphic schemes based on the brand content, industry and color preference information input by the user, and supports online editing, series VI scene display and business card design.

[0018] Furthermore, the intelligent logo design system also includes:

[0019] Deep learning image classification technology, deep learning training evaluation network and massive trademark image similarity retrieval technology based on deep learning convolution features to achieve rapid transformation, fusion, generation and retrieval of logo elements.

[0020] Furthermore, the intelligent poster design system includes:

[0021] The automated design of posters is achieved by establishing a material center, planning and designing a framework for deep reinforcement learning training models, training and evaluating networks to ensure the quality of poster design, and the generator rendering process.

[0022] Furthermore, the intelligent cutout system comprises:

[0023] The original image and the three-dimensional image are simultaneously input into the neural network model, and the image features are extracted with the help of the convolutional network, which is the encoding process;

[0024] Use transposed convolution to improve resolution and predict an alpha channel of the same size as the input, which is the decoding process;

[0025] The entire encoding and decoding process constitutes the first stage of the neural network. The second stage is the refinement stage. After merging the original image and the predicted alpha channel, four convolution operations are performed to output the final alpha channel prediction value.

[0026] Furthermore, the intellectual property management system includes:

[0027] Combining AI technology with the field of intellectual property, it provides a variety of functions including online trademark management, trademark monitoring, trademark classification query, trademark search and trademark intelligent registration, and supports the entry of agents and the integration of online and offline services.

[0028] Furthermore, the one-stop service platform adopts a high-performance Solr search engine based on a Lucene-based full-text search server.

[0029] The present invention also provides a method for applying a one-stop service platform, using any of the one-stop service platforms described above to achieve automated and intelligent services for small and medium-sized enterprises in naming, design, data analysis and intellectual property management.

[0030] The beneficial effects of the present invention include at least:

[0031] (1) The platform significantly improves the brand creation and market response speed of SMEs by integrating intelligent naming and design functions;

[0032] (2) By leveraging deep learning and big data technologies, the platform can reduce trademark registration risks and increase registration success rates;

[0033] (3) Intelligent design functions reduce human intervention, shorten design cycles, and reduce costs while maintaining the originality and innovation of the design;

[0034] (4) The one-stop service model simplifies the enterprise's operational processes in naming, design, data analysis and intellectual property management;

[0035] (5) The platform’s automated and intelligent services have improved the market competitiveness of small and medium-sized enterprises and lowered the threshold for professional technical support. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A schematic diagram of the architecture of an intelligent naming system on a one-stop service platform based on artificial intelligence provided by the present invention.

[0037] Figure 2 A schematic diagram of the architecture of an intelligent logo design system on a one-stop service platform based on artificial intelligence provided by the present invention.

[0038] Figure 3 A logo design flow chart of an intelligent logo design system on a one-stop service platform based on artificial intelligence provided by the present invention.

[0039] Figure 4 A schematic diagram of the overall process of an intellectual property management system on a one-stop service platform based on artificial intelligence provided by the present invention.

[0040] Figure 5 A schematic diagram of the architecture of an intellectual property management system on a one-stop service platform based on artificial intelligence provided by the present invention. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The present invention discloses a one-stop service platform based on artificial intelligence, which integrates at least:

[0043] Intelligent naming system: By using LSTM model and big data, combined with industry information and / or legal person information, online naming is realized;

[0044] Intelligent logo design system: Through NLP and sequence planning technology, it can quickly generate and edit logo solutions based on brand information and preferences, including generating VI scene displays and business card designs;

[0045] Intelligent poster design system: using the established material center and deep reinforcement learning to automatically design personalized posters;

[0046] Intelligent cutout system: uses convolutional neural network to predict the alpha channel of the image from the original image and the three-dimensional image to achieve accurate cutout of the image;

[0047] Intellectual Property Management System: Combined with AI, it provides comprehensive online intellectual property services, including management, monitoring and registration.

[0048] like Figure 1 As shown: the intelligent naming system includes:

[0049] It adopts a deep learning LSTM neural network model, links at least 90 million industrial and commercial enterprise information databases and at least 300,000 Tang poetry and Song Dynasty lyrics libraries, and uses word embedding technology and deep learning LSTM network training models to achieve online naming. According to the industry information of the company entered by the user, hundreds of company names can be generated online for selection. It can also support additional input of company legal person information to generate candidate company names.

[0050] Generating candidate company names through an intelligent naming system is fast, convenient and has a high registration approval rate.

[0051] Furthermore, the intelligent naming system also includes:

[0052] Based on the big data of registered corporate identities (i.e. company names and corporate trademarks) and information on companies to be registered, multi-dimensional analysis using artificial intelligence algorithms and industrial and commercial registration comparison and duplication testing are used to generate candidate company names with batch and high industrial and commercial registration pass rates.

[0053] This system uses a high-performance Solr search engine based on the Lucene full-text search server, which indexes and stores 96 million company registration data, 43 million trademark data and 1 million sensitive vocabulary. The database is regularly updated and maintained, and an inverted index (reverse index) is used to achieve millisecond-level search speeds.

[0054] like Figure 2-3 As shown, the intelligent logo design system includes:

[0055] Using deep learning natural language processing technology and deep learning sequence planning network model technology, it generates multiple sets of alternative logo graphic schemes based on the brand content, industry and color preference information input by the user, and supports online editing, series VI scene display and business card design.

[0056] Furthermore, the intelligent logo design system also includes:

[0057] Deep learning image classification technology, deep learning training evaluation network and massive trademark image similarity retrieval technology based on deep learning convolution features to achieve rapid transformation, fusion, generation and retrieval of logo elements.

[0058] Specifically, the intelligent logo design system will generate a large number of logo graphic solutions for selection within 30 seconds based on the brand content (name, subtitle), industry and color preference information entered by the user. The system supports online editing for each logo graphic, and users can optimize and adjust it in the direction they like. In addition, for each logo solution, the system will generate a series of VI scene displays and business cards and other extension designs.

[0059] The intelligent logo design system comprises the following steps:

[0060] ① Deep learning natural language processing technology

[0061] The deep learning RNN recurrent neural network is used to parse the text information input by the user and extract the effective information. The effective information is represented as a word vector through the word2vec algorithm, and the text is digitized to make it easier for the computer to understand. On this basis, the platform obtains key information through the text content input by the user. Based on this key information, the platform can better integrate the user's brand information when designing the logo online.

[0062] ②Output LOGO design framework based on training deep learning sequence planning network model technology

[0063] The labeled LOGO design data is input into the deep learning sequence network, and after learning through this layer of neural network, a design framework is obtained.

[0064] A major feature of deep learning sequence networks is that they have a certain memory function. Because design is a complex process, it often takes dozens of steps to complete a design. Technically speaking, it is a model composed of a bunch of spatial features and visual features. From the perspective of a designer, it is equivalent to the designer's rough framework impression before making a set of designs. For example, today a designer receives a task to make a Chinese-style logo, thinks about how this design will be done, and then extracts a feature model from a bunch of files.

[0065] Based on the user's input of brand name, brand slogan, favorite color and industry information, the deep learning natural language processing (NLP) technology is used to effectively extract key information from the user's input information, and the basic materials in the database are effectively transformed and integrated with image algorithm technology to quickly generate logos in batches; through the similarity fast retrieval technology of massive trademark images, the effective logo is output and displayed to the front-end user; the product can generate unlimited logo solutions for users within 30 seconds. To a certain extent, it has subverted the traditional inherent mode of users seeking designers for logo design, greatly improving work efficiency.

[0066] ③Deep learning image classification technology

[0067] Smart design is based on element-level design, so an element library must be prepared. By collecting some copyrighted image libraries and original design elements, they are input into an element classification model. This classification model will classify these elements into various types, such as background, main visual elements, and auxiliary elements, and finally establish an element center with classification attributes.

[0068] The engine selects the design framework based on the needs and historical user data analysis, selects elements from the element center, and traverses the state space.

[0069] ④ Deep learning training evaluation network to score LOGO quality

[0070] The scoring model is trained by manually inputting a large number of existing logo design drawings, design rules and scoring data. At the same time, experts will also manually intervene in the scoring to complete two-way feedback. For the finished logo designed, the evaluation network model is input and the corresponding quality score is output.

[0071] ⑤ Similarity retrieval technology for massive trademark images based on deep learning convolutional features

[0072] Based on convolutional neural networks, the platform extracts deep learning-based image features from 30 million trademark data. By analyzing the correlation between the features themselves and using the principal component analysis algorithm, the feature dimension is reduced while minimizing information loss. On this basis, the features are hash-coded and used for fast image retrieval. Accurate and efficient retrieval results have been achieved. The average retrieval accuracy reaches 99.2%.

[0073] ⑥ Rapid name query technology based on search engine

[0074] It uses advanced solr search engine technology and combines it with the efficient mongodb database to realize rapid search of massive data. It can achieve real-time, multi-dimensional, and multi-precision queries, and supports priority sorting, result aggregation, dynamic batch update and other functions.

[0075] The poster design process is an orderly superposition process of a series of material layers with certain characteristic attributes. The characteristic attributes here include both visual attributes such as color, shape, texture, theme, etc., as well as spatial attributes such as position, size, and edge.

[0076] The intelligent poster design system comprises:

[0077] 1) Establish a material center: complete the digitization of materials

[0078] Based on traditional machine learning methods, low-level semantic (color, shape, etc.) features of material images are extracted, and deep learning technology is combined to extract high-level semantic features to jointly complete the extraction of material feature attributes.

[0079] 2) Planner: A deep reinforcement learning training model planning and design framework

[0080] The poster design process is a process of stacking material layers in sequence. This process is fitted by a deep sequence generation model. In the modeling, the material is regarded as a word, the poster is regarded as a sentence, the vocabulary index is regarded as a discretized feature sequence, and the material stacking order is regarded as the word order in the sentence.

[0081] 3) Optimizer: Training and evaluating the network to ensure the quality of poster design

[0082] Design some objective functions of basic conventional design concepts and aesthetic standards, and the action set includes moving, scaling, brightness adjustment, etc. Based on the designer team scoring the pre-given poster data, obtain the scoring data, train the evaluation model, and improve the visual effect of the smart poster.

[0083] 4) Generator: The generator renders the optimized material feature sequence into a graph through the planned design framework.

[0084] The planner learns from the data the designer's design habits and rules for different styles. The optimizer makes fine adjustments to the output results of the former based on aesthetic quality and design principles. The generator selects / generates materials and renders them into images. The material center serves as the foundation and is responsible for material management and material labeling.

[0085] The intelligent cutout system comprises:

[0086] The original image and the triplicated image are simultaneously input into the neural network model, and the image features are extracted with the help of the convolutional network, i.e., the encoding process. The triplicated image is an image that is marked into three parts: a determined foreground target area, a determined background area, and the remaining uncertain area.

[0087] Use transposed convolution to improve resolution and predict an alpha channel of the same size as the input, which is the decoding process;

[0088] The entire encoding and decoding process constitutes the first stage of the neural network. The second stage is the refinement stage. After merging the original image and the predicted alpha channel, four convolution operations are performed to output the final alpha channel prediction value.

[0089] In the traditional design service model, frequent communication and modification between the demander and the designer are required to reach a satisfactory design solution, which takes a long time and is difficult to assess the risk of graphic registration. The auxiliary design system based on deep learning graphic generation and retrieval technology can quickly generate a large number of design solutions with low trademark registration risks for reference by the demander, which can effectively improve the service efficiency of the industry.

[0090] In addition, the auxiliary design system requires a large number of original basic graphic elements, and manual design alone is inefficient. Generating a large number of variable basic element graphics through deep learning generative adversarial networks can efficiently generate the required graphic data, reducing the reliance of the auxiliary design system on human designers.

[0091] Furthermore, the intellectual property management system includes:

[0092] Combining AI technology with the field of intellectual property, it provides a variety of functions including online trademark management, trademark monitoring, trademark classification query, trademark search and trademark intelligent registration, and supports the entry of agents and the integration of online and offline services.

[0093] Specifically, Figure 4-5 As shown in the figure, AI technology is effectively combined with the field of intellectual property to create an online intellectual property management service platform based on big data that links users and professional agents, providing highly professional and easy-to-use intellectual property services for trademark practitioners and related institutional clients. By recruiting agents to join, offline users are guided online, which promotes the profitability of other products on the platform, and connects all corporate service product lines of Yipinweike.com to form a closed loop. The intellectual property management platform integrates trademark big data artificial intelligence technology, realizes online trademark management, trademark monitoring, trademark classification query, trademark search, trademark intelligent registration and other functions, which greatly facilitates user use.

[0094] Furthermore, the one-stop service platform adopts a high-performance Solr search engine based on a Lucene-based full-text search server.

[0095] The present invention also provides a method for applying a one-stop service platform, using any of the one-stop service platforms described above to achieve automated and intelligent services for small and medium-sized enterprises in naming, design, data analysis and intellectual property management.

[0096] The platform is based on the standard Restful architecture and the lightweight and scalable Flask microservice framework, providing unified services for standard front-end / application back-end / Mobile Web / Mobile App / third-party platforms.

[0097] The one-stop service platform proposed in this invention provides a comprehensive solution for small and medium-sized enterprises by deeply integrating "big data + artificial intelligence" technology, covering a full range of services from naming, design, data analysis to intellectual property management. The advantage of this platform lies in its highly integrated and automated characteristics, which can significantly improve the efficiency and effectiveness of enterprises in brand building, market response speed and intellectual property protection.

[0098] First, the smart naming function uses the LSTM neural network model, combined with a huge industrial and commercial enterprise information database and traditional cultural resources, to quickly generate a company name that conforms to industry characteristics and cultural traditions, greatly shortening the company naming cycle and increasing the registration approval rate of the name. The smart logo design and smart poster design functions use deep learning and natural language processing technology to quickly generate design drafts based on the user's specific needs, and adjust them through online editing tools, greatly improving the flexibility and response speed of the design.

[0099] Intelligent cutout technology uses convolutional neural networks to accurately predict the alpha channel of an image, making image processing more efficient and accurate, providing companies with high-quality visual materials. The intellectual property management platform combines AI technology to provide online trademark management, monitoring, query and registration services to help companies effectively manage and protect their intellectual property and reduce legal risks. This platform not only improves work efficiency and reduces costs, but also improves design quality and the professionalism of intellectual property protection through intelligent services, providing strong technical support for the rapid development of small and medium-sized enterprises.

[0100] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A one-stop service platform based on artificial intelligence, characterized by: At least integrated: Intelligent naming system: By using LSTM model and big data, combined with industry information and / or legal person information, online naming is realized; Intelligent logo design system: Through NLP and sequence planning technology, it can quickly generate and edit logo solutions based on brand information and preferences, including generating VI scene displays and business card designs; Intelligent poster design system: using the established material center and deep reinforcement learning to automatically design personalized posters; Intelligent cutout system: uses convolutional neural network to predict the alpha channel of the image from the original image and the three-dimensional image to achieve accurate cutout of the image; Intellectual Property Management System: Combined with AI, it provides comprehensive online intellectual property services, including management, monitoring and registration.

2. The one-stop service platform based on artificial intelligence according to claim 1, characterized in that: The intelligent naming system comprises: A deep learning LSTM neural network model is used to link at least 90 million industrial and commercial enterprise information databases and at least 300,000 Tang and Song poetry libraries, and candidate company names are generated based on the company's industry information and / or company legal person information entered by the user.

3. The one-stop service platform based on artificial intelligence according to claim 2, characterized in that: The intelligent naming system also includes: Based on the multi-dimensional analysis of the big data of registered enterprise logos and information of enterprises to be registered, as well as the comparison and duplication test of industrial and commercial registration, candidate enterprise names with batch and high industrial and commercial registration pass rates are generated.

4. The one-stop service platform based on artificial intelligence according to claim 1, characterized in that: The intelligent logo design system includes: Using deep learning natural language processing technology and deep learning sequence planning network model technology, it generates multiple sets of alternative logo graphic schemes based on the brand content, industry and color preference information input by the user, and supports online editing, series VI scene display and business card design.

5. The one-stop service platform based on artificial intelligence according to claim 4, characterized in that: The intelligent logo design system further includes: Deep learning image classification technology, deep learning training evaluation network and massive trademark image similarity retrieval technology based on deep learning convolution features to achieve rapid transformation, fusion, generation and retrieval of logo elements.

6. The one-stop service platform based on artificial intelligence according to claim 1, characterized in that: The intelligent poster design system comprises: The automated design of posters is achieved by establishing a material center, planning and designing a framework for deep reinforcement learning training models, training and evaluating networks to ensure the quality of poster design, and the generator rendering process.

7. The one-stop service platform based on artificial intelligence according to claim 1, characterized in that: The intelligent cutout system comprises: The original image and the three-dimensional image are simultaneously input into the neural network model, and the image features are extracted with the help of the convolutional network, which is the encoding process; Use transposed convolution to improve resolution and predict an alpha channel of the same size as the input, which is the decoding process; The entire encoding and decoding process constitutes the first stage of the neural network. The second stage is the refinement stage. After merging the original image and the predicted alpha channel, four convolution operations are performed to output the final alpha channel prediction value.

8. The one-stop service platform based on artificial intelligence according to claim 1, characterized in that: The intellectual property management system includes: Combining AI technology with the field of intellectual property, it provides a variety of functions including online trademark management, trademark monitoring, trademark classification query, trademark search and trademark intelligent registration, and supports the entry of agents and the integration of online and offline services.

9. The one-stop service platform based on artificial intelligence according to claim 1, characterized in that: The one-stop service platform adopts a high-performance Solr search engine based on a Lucene full-text search server.

10. A method for applying a one-stop service platform, characterized in that: Use the one-stop service platform described in any one of claims 1-9 to provide automated and intelligent services for small and medium-sized enterprises in naming, design, data analysis and intellectual property management.

Images