Database construction and remodeling method on traditional Yi nationality patterns of China with GAN model

Abstract

Traditional Yi ethnic patterns, as vital cultural heritage, increasingly face marginalization in modern design contexts due to regional cultural specificity and shifting esthetic demands. This study pioneers a GAN-driven framework to digitally preserve and innovatively regenerate these patterns. A hybrid approach combining manual curation with computational preprocessing (e.g., eigenvector analysis, semantic annotation) established a structured database of Yi motifs. Leveraging this repository, a customized adversarial generative network was developed to synthesize culturally authentic yet stylistically adaptive patterns. Rigorous evaluations confirmed the model’s capacity to balance ethnic identity retention with creative diversification. The generated patterns demonstrate practical applicability in textile, architectural ornamentation, and digital media, bridging heritage conservation with contemporary design needs. By integrating database construction, generative modeling, and application-oriented validation, this work provides a scalable paradigm for revitalizing endangered traditional patterns across ethnic communities, offering both technical methodologies and cross-disciplinary insights for cultural sustainability.

Introduction

As vital embodiments of a nation’s cultural heritage, traditional ethnic patterns offer invaluable insights into national and global history. However, influenced by contemporary design and commercial esthetics, the unique regional features of these patterns have, in practice, hindered their wider diffusion and assimilation across regions. Moreover, modern design, with its emphasis on simplicity, efficiency, and utility, stands in contrast to the complex and timeless nature of traditional patterns, progressively relegating them to the periphery of dissemination and commercial use.

The existing literature on traditional patterns worldwide is diverse, primarily concentrating on ontology, connotation, and practical applications. Certain researchers evaluate the symbolism in the patterns themselves, assessing their potential employs in contemporary design^1,2,3,4,5. Scholars such as Li Zhiwei and Hu Bingyu have categorized and systematized traditional Chinese patterns, integrating them with modern design principles to develop a digital preservation platform^1,2,3. Ghazal Refalian et al. analyzed the formal characteristics of Islamic geometric patterns⁴. Hari Nugraha et al. assessed the visual impact of traditional Indonesian patterns applied to bamboo⁵. While these studies are commendable in their efforts to protect and conserve traditional patterns, further analysis is needed into how these patterns can achieve widespread integration into contemporary products.

Recently, fueled by the rapid advancements in digital technology and artificial intelligence, research on traditional ethnic patterns has progressively transitioned from a solely cultural or artistic focus toward interdisciplinary synthesis and creative development. In modern design applications, a growing body of research explores how digital technologies (such as GAN models, parametric design tools, etc.) can reinterpret patterns and imbue them with more varied forms of expression. In addition, the incorporation of traditional patterns into contemporary fashion, interior design, and product packaging has steadily grown, establishing a research direction that emphasizes concurrent preservation and innovation. This trend not only cultivates the dissemination of ethnic patterns in global cultural exchange but also creates ample opportunities for their utilization in modern commercial markets.

The Yi nationality constitutes the sixth largest ethnic minority group in China, distributed primarily across the provinces of Sichuan, Yunnan, and Guizhou. Their culture, reflected by depth and richness, demonstrates diverse forms of expression. As a core and characteristic of China’s regional cultural culture, Yi culture has long played an important role in preserving and advancing China’s exceptional traditional cultural heritage^6,7. Liangshan, located in Sichuan province, represents the largest Yi community in China, with a rich history and a culture produced by numerous significant events. Traditional Yi patterns are the social product of the development of the Yi people to a certain extent and are the essence of folk culture. Further exploration of their cultural significance, perpetuation of their esthetic merit, optimization of their underlying cultural genes, and application of digital redesign hold significance for the revitalization and renewed application of cultural resources in the cultural industry. Yi patterns represent a core component of Yi culture. Designated as a national intangible cultural heritage of China, these patterns exhibit high artistic, cultural, and technical value. Their applications extend across various domains, including Yi attire and accessories, daily objects, and decorative arts. The underlying culture, regional characteristics, forms, and colors in these patterns carry rich cultural and religious connotations and are significant for inheritance. It can be argued that the innovative redesign of traditional Yi patterns prioritizes the preservation of their cultural meanings over the mere replication of the symbols themselves^8,9,10.

Notwithstanding China’s rapid modernization and urbanization, Yi nationality patterns are gradually vanishing, and their preservation faces numerous obstacles⁸. Over recent decades, researchers analyzing the preservation crisis of Yi nationality patterns have identified several critical issues, including market operational inefficiencies, technological stagnation, and mismatched supply and demand, then offering recommendations to local cultural contexts and market conditions^11,12,13,14. While existing research on Yi cultural preservation primarily evaluates patterns through artistic, cultural transmission, and historical perspectives, limited research addresses the technological and smart applications for pattern inheritance and product innovation. Based on available literature, the current approaches to intelligent transformation and innovative inheritance of traditional Yi patterns can be categorized into four main aspects:

• The first aspect centers on analyzing the cultural significance of Yi nationality patterns. Research in this domain studies the regional cultural elements, humanistic characteristics, and symbolic meanings of these designs. For instance, Chen Yumeng’s research evaluates the inheritance and innovation of Yi patterns in lacquerware, considering their forms, applications, and types¹⁵. Yang Yuping’s work evaluates the artistic features and esthetic significance of Yi embroidery patterns while exploring their cultural implications and symbolic importance¹⁶. Comprehensive information about Yi nationality patterns can be discovered in publications such as Research Series on Ancient Esthetic Thoughts of Chinese Ethnic Minorities and Research on Esthetic Thoughts of Chinese Ethnic Minorities.

• A second area of research focuses on constructing pattern databases. Researchers employ artificial intelligence and computer technology to store, retrieve, and process pattern data, thus supporting the inheritance and development of traditional culture^17,18. For instance, patterns are first collected and a pattern gene bank is created utilizing web crawler technology, parameterized model development, and visual saliency algorithms¹⁹. Then, the generative design of pattern genes is completed through integration with mass customization technology. Finally, a relational model is implemented to maintain data integrity and security, and the database is systematically and scientifically deployed for pattern preservation and storage²⁰.

• A third research direction centers on reimagining traditional patterns through image-generation model methods. Research teams and technical designers use software to produce artistic texture images, which are then repaired, reconstructed, compressed, and interpreted^21,22. For instance, they draw upon style transfer, image enhancement, and recognition methods for animation, fashion, and layout design^23,24. Alternatively, they employ image generation technology to develop an intelligent digital platform for data design, analysis, and dissemination²⁵. Image generation and training methods are applied to repair, reconstruct, compress, and interpret specific images²⁶.

• A fourth research area evaluates the innovative design and application of patterns. Functional studies and intelligent analyses of patterns are undertaken to develop diverse changing, creative, and broad contemporary design methodologies, drawing primarily from the natural and social sciences. Simultaneously, these methodologies are applied to media such as ethnic clothing, artifacts, and architecture, and integrated with new technological approaches to broaden their application^27,28.

Current research trends and dynamics across the four aforementioned areas indicate several limitations: inadequate integration of traditional pattern development with modern technology; challenges in collecting and preprocessing pattern data; an inability to effectively identify and extract characteristic pattern styles; and a lack of image generation models capable of supporting complex patterns. Simultaneously, the theories presented in existing research seldom address practical strategies for integrating traditional ethnic patterns into contemporary commercial markets. This has contributed to the continued marginalization and decline of Yi nationality patterns in modern design.

Through repeated practical application and analysis; however, it has been observed that Generative Adversarial Networks models offer robust generative capabilities, exceptional self-learning and optimization potential, efficient and accurate results, and significant adaptability to complex data. These characteristics are beneficial for creating and generating pattern forms appropriate for contemporary design applications. In contrast to traditional manual design or rule-based algorithmic generation methods, GAN models can, by learning from extensive existing ethnic pattern datasets, deeply explore the cultural characteristics and design logic behind patterns. They can also produce innovative, novel pattern styles while preserving the fundamental qualities of traditional patterns. Critically, GAN models offer adaptive optimization capabilities, enabling continuous adjustment of generation parameters based on specific design needs, thus improving the accuracy and applicability of the generated results. Design efficiency and visual innovation are critical factors of product competitiveness, particularly in contemporary commercial markets. GAN models can considerably reduce pattern design development time while simultaneously creating visually striking and culturally relevant patterns, offering brands and designers more creative choices. This technology connects traditional ethnic art with modern design, preserving the unique allure of ethnic culture while reviving its innovation and dissemination in globalization. In addition, the author Kaining Meng notes that the working method demonstrated by the GAN model has “universality” in the development of traditional ethnic patterns. That is, the methods and results demonstrated by the GAN model in the experiment of reshaping Yi nationality patterns can be widely applied to the practice of other related ethnic patterns.

This study produces a method for reinterpreting traditional ethnic patterns based on GAN model application. Leveraging the operational characteristics of GANs, a novel four-stage workflow (esthetic analysis—database establishment—developing GAN model—application and evaluation) will be presented. Focusing on Yi nationality patterns, this research will reinterpret and regenerate these patterns. It is hoped that this study’s method will offer broadly applicable guidance for the reinterpretation of traditional patterns among similar ethnic groups. Moreover, it is expected that this research will suggest new avenues and help with the preservation, transmission, reinterpretation, and innovative application of traditional patterns globally.

In designing and implementing the proposed method, this study will address limitations identified in the literature review, pursuing innovation in Yi patterns through the integration of “technology” and “art.” Technologically, this involves utilizing digital image processing and synthesis to construct Yi pattern databases and image generation models. This approach also supports the storage, preservation, transmission, and reinterpretation of Yi patterns. Artistically, this involves analyzing the cultural meanings and historical development in these patterns. This analysis comprises the optimization, reconstruction, and deconstruction of patterns, helping their appropriate transformation and application in diverse design contexts to achieve pattern redesign and innovation.

Methods

Building process from image database to GAN modeling

Central to this study is a method for transforming traditional patterns (Fig. 1). Later research will explore the implementation and applications of this transformation method.

Fig. 1

Research process’s four sections.

This research method and process consists of four parts. The following sections will detail the experimental methods and key technologies employed in each part.

The workflow is divided into the following four steps:

• Analysis of esthetic style of traditional patterns: This study conducts an in-depth analysis of the esthetic styles of traditional Yi ethnic patterns, focusing on color, form, and application vectors. This foundational analysis offers theoretical insights and design guidance for digital preservation and contemporary applications.

• Database establishment for traditional patterns: A comprehensive database of traditional Yi patterns is developed through pattern collection, feature extraction, and image annotation. This database forms the foundation for model training and creative pattern design.

• Developing the pattern-generating model: This study trains a generative adversarial network (GAN) to synthesize pattern images, optimizing the generator and discriminator to enhance quality and diversity. The GAN model, central to the pattern-generation framework, comprises two key components: a generator that creates new patterns and a discriminator that evaluates their similarity to authentic Yi patterns. During training, images from the traditional pattern database teach the generator to produce novel patterns with traditional features, while the discriminator learns to differentiate real patterns from generated ones. As training progresses, the generator utilizes learned features to produce innovative patterns, a pivotal step in reshaping traditional designs. The discriminator is refined to better distinguish real patterns from generated ones, prompting the generator to create higher-quality designs. Generator refinement is guided by feedback from the discriminator, involving adjustments to architecture or training parameters to improve quality and diversity. The refined generator ultimately synthesizes pattern images, used in evaluation and application phases to validate the model’s effectiveness and innovation. This framework reimagines traditional Yi patterns, supporting cultural preservation while offering technical guidance and inspiration for modern product design.

• Application and evaluation of patterns: Patterns generated by the GAN are evaluated for innovation and practical applications.

Analysis of esthetic style of traditional patterns

Foundational in this research is an analysis of art style. This analysis develops a comprehensive understanding of the visual language, design principles, and cultural significance of traditional ethnic patterns. This understanding offers accurate and effective data, guiding the deep learning processes of the GAN models. By exploring the unique visual characteristics and cultural meanings of ethnic patterns, this analysis offers clear identification of pattern characteristics for training the GAN models, ensuring that generated patterns inherit traditional esthetic qualities while preserving cultural integrity. This process creates a high-quality training dataset, enabling the model to effectively identify and reconstruct the visual characteristics of various patterns. The key design elements identified through this analysis also act as critical parameters in the pattern generation stage, optimizing the detail quality and overall coherence. The following section will illustrate this analytical phase by focusing on “Yi nationality patterns” as a case study, detailing the methodological approach.

Yi culture, influenced by natural forces, scientific and technological advancements, societal structures, economic progress, and social mores, has evolved over a long history. This study, drawing upon a three-tiered theory of culture, employs extensive fieldwork and archival research to appraise traditional Yi patterns. These patterns are analyzed across material, functional/technological, and spiritual/esthetic dimensions to thoroughly explore their cultural significance and artistic merit.

The visual and concrete carriers of Yi culture are deep in the material stratum. At the behavioral level, underlying cultural principles and their transformations are reflected in the conventions and practices connected with the material stratum. The spiritual/esthetic level forms the bedrock and foundation for cultural continuity. This necessitates further research of the material, behavioral, and spiritual/esthetic dimensions of traditional Yi patterns, dividing them into various cultural characteristics, and categorizing their cultural expressions and esthetic qualities, as detailed in Table 1.

Table 1 Analysis of esthetic style of Yi traditional patterns

The outward expression of cultural values is clearly apparent in the material stratum, which exhibits a rich array of cultural symbols through appropriate visual forms. This study mainly evaluates the characteristics of Yi patterns across three dimensions: type, color, and composition. In terms of pattern types, Yi artisans employ methods of exaggeration and deformation while maintaining the essential forms of their subjects. They express their aspirations for a better life through patterns ranging from abstract and irregular, to concrete representations of nature such as flora, fauna, and insects, and finally to regular geometric designs. Regarding color, the Yi people of Liangshan primarily utilize the main colors of red, yellow, and black, coordinating these with secondary hues. Compositionally, Yi artists abstract the principles of nature to express an esthetic of order. They adhere to principles of variety, unity, balance, proportion, scale, rhythm, and rhyme. Moreover, they utilize methods such as repetition, continuity, symmetry, equilibrium, and hierarchical organization to create patterns that are both complex and well-ordered. Common compositional structures include several designs, radiating patterns, and bilateral symmetry^9,10,29.

In their practical and technological dimensions, Yi nationality patterns constitute a social practice expressed through production methods, lifestyles, and customs shaped by specific environments. This study evaluates the folk character of these patterns, analyzing them through the perspective of application scenarios and production methods. Regarding application scenarios, Yi patterns appear frequently in traditional architecture, furniture, and craft objects such as lacquerware, silverware, apparel, carvings, and paintings. Their configurations are constrained by factors such as production methods, available materials, and artistic intent. As for production methods, Yi apparel typically incorporates embroidery, floral artistry, and painted decoration. Embroidery is frequently employed in blouses, headscarves, small bags, and women’s skirts, while floral artistry comprises methods including picking, pasting, piercing, trifling, appliqué, and coiling. Painting, accordingly, comprises three basic components: dots, lines, and planes. Yi crafts and methods demonstrate programmatic characteristics that can exhibit order and rhythm³⁰.

The belief and esthetic dimensions of these patterns offer insights into the psychology and humanistic values of the Yi people. This study summarizes their ethnic beliefs, esthetic principles, and underlying philosophies of creation. The beliefs of the Yi people generally center around the veneration of nature, ancestors, and symbolic representation. Their esthetic preferences often draw inspiration from nature and work, giving voice to aspirations for a wonderful and united existence. In their creative philosophy, the Yi people prioritize utility, ethnic identity, religious significance, and a primal esthetic sensibility.

Analyzing the esthetic qualities of Yi patterns through the method of esthetic logic can render the cultural phenomena of traditional Yi nationality patterns more regular and give a theoretical basis for the digital transformation of patterns. Digitally reimagining traditional Yi patterns necessitates exploring material properties, integrating contemporary values, and welcoming innovative components. It also demands adherence to the cultural core, assimilating valuable advancements from broader societal development, and changing the ethnic cultural meaning and artistic character into a design language compatible with computer-based artificial intelligence.

Furthermore, it is worth noting that the construction of the GAN model primarily depends on a detailed analysis of tangible pattern features, including morphology, color, and medium. In contrast, analyzing abstract elements like beliefs and philosophies offers unique academic value. Such analysis helps researchers systematically classify and manage pattern data during collection and offers a multidimensional retrieval system for later database use. For instance, researchers can search the database by visual attributes like color and morphology or by abstract keywords such as symbolic meaning, implications, or religious beliefs. This approach improves database clarity, enabling efficient matching with specific research needs.

Meanwhile, a key goal of reshaping patterns with the GAN model is to integrate generated patterns into modern product design, enhancing their market potential. Summarizing the abstract meanings of generated patterns provides valuable references for pattern applications. For instance, product designers can choose patterns based on visual features or their cultural connotations, emotional resonance, and social symbolism. Distilling and summarizing abstract meanings enhance the cultural value and usability of generated patterns while offering theoretical and practical guidance for applying traditional patterns in modern contexts. Ultimately, this research provides new pathways for innovatively inheriting and commercializing traditional patterns.

Following an analysis of the patterns’ esthetic qualities, the next phase involves developing a GAN model-based pattern database, organized according to these esthetic principles. This process comprises six key steps: manual image data collection, image pre-processing, calculation of pattern eigenvectors, computational data acquisition, pattern annotation, and adjustment and improvement of the database. Utilizing Yi nationality patterns as a case study, this research will continue to take Yi nationality patterns as an example to introduce the specific operational methods for each step.

First, manual data collection is essential for database construction. Once a sufficient sample size is achieved, eigenvector extraction is performed on these samples to enable computerized database expansion. Finally, the database is improved through numbering and manual adjustment. This lays the foundation for the construction of image-generation models. This stage applies digital acquisition, processing, repair, preservation, management, integration, and other means to store, acquire, and process pattern data through artificial intelligence and computer programs³¹. It builds a database based on this to achieve digital inheritance and protection of traditional culture. The entire process is divided into the following six steps, as presented in Fig. 2:

• Step 1—Manual acquisition of patterns: First, pattern data are collected from websites featuring Yi traditional culture, with promotional materials, e-commerce platforms, cultural and creative products, apparel, and historical artifacts. This dataset, exceeding 1000 pieces, produces a foundation for image pre-processing and computerized image acquisition.

• Step 2—Image pre-processing: Image pre-processing is essential to enhance image quality and recover valuable authentic information potentially compromised by varying degrees of noise introduced during image acquisition. This procedure involves the following three steps. Specifically, grayscale or binary conversion is omitted for the patterns, as the color of Yi nationality patterns is a crucial medium for conveying ethnic beliefs and meanings.

  Fig. 2

  

  The database construction of Yi traditional pattern.

  Filter pre-processing: Common filters include the mean filter, median filter, and Gaussian filter. Considering the complexity of the patterns and the Gaussian filter’s beneficial properties, such as its effective smoothing, edge preservation, and adaptability, it is selected for pattern pre-processing³².

  Denoising pre-processing: Established methods include wavelet denoising, BM3D denoising, and NL means denoising. Wavelet denoising is preferred for pre-processing pattern images due to its superior performance in preserving image detail, frequency selectivity, adaptability, and broad applicability³³.

  Normalization pre-processing: This critical step adjusts image brightness, contrast, and color balance, and scales pixel values to a specified range, optimizing them for algorithmic requirements³⁴.

  $$H(i,j)=1/(2\pi {\sigma }^{2})* \exp (-({(i-m)}^{2}+{(j-n)}^{2})/(2{\sigma }^{2}))$$

  (1)
  $${X}^{\wedge}={W}^{\wedge}\,\cdot\,X$$

  (2)
  $$I^{\prime\prime} (x^{\prime} ,y^{\prime} )=I^{\prime} (x^{\prime} ,y^{\prime} )-{I}\_{\min }/{I}\_{\max }-{I}\_{\min }$$

  (3)

  The calculations for the three preceding pre-processing steps are given by formulas (1)–(3), respectively. In formula (1), “H(i, j)” denotes the elements of the convolution kernel, “m” and “n” represent the kernel’s center coordinates, and “σ” is the Gaussian function’s standard deviation. In formula (2), “W^” denotes the inverse wavelet transform matrix, and “X^“ represents the denoised image. In formula (3), “I″”and “I′’” represent the pixel values of the normalized and adjusted images, respectively, while “I_min” and “I_max” denote the corresponding minimum and maximum pixel values. In short, pre-processing pattern images in MATLAB, OpenCV, and Adobe Photoshop utilizing this procedure efficiently creates a preliminary database of Yi nationality patterns derived from manually collected samples.

• Step 3—Pattern eigenvector calculation: To expand the dataset through automated online image collection, it is essential to verify whether these online images indeed represent Yi nationality patterns. The eigenvector calculation, involving geometric and textural features, serves as a critical verification step, ensuring that the constructed pattern database accurately reflects the characteristics of traditional Yi patterns. This process enhances the precision of subsequent classification, recognition, and image-matching tasks.

  The technical implementation of this step is structured into three clearly defined phases:

  Feature retrieval: Feature retrieval encompasses both geometric and textural aspects. Geometric features primarily include pattern dimensions, contours, and edge characteristics, which are extracted via edge detection algorithms (e.g., Canny edge detection) and contour detection algorithms. Textural features such as color distribution, grayscale variations, and texture patterns are acquired through Gaussian filtering and wavelet transforms. Specifically, Gaussian filtering reduces noise and preserves edge clarity, which is crucial for accurate edge and contour detection. Its mathematical expression is provided as follows³⁵:

  $$G(x,y)={1}/({2\pi {\sigma }^{2}})\,*\,{{{exp}}}\,{(-({{x}^{2}+{y}^{2}})/({2{\sigma }^{2}))}}$$

  (4)

  where “G (x, y)” denotes the two-dimensional Gaussian filter function and represents the standard deviation. Gaussian filtering efficiently removes image noise while preserving crucial edge details. The wavelet transform is mathematically expressed as

  $$W=\mathop{\sum}\limits_{n}f(n)\,\psi(n)$$

  (5)

  where “f(n)” is the original image data at sampling point “n”, “ψ(n)” represents the wavelet transform function, and signifies the wavelet coefficient values.

  In practical implementation, we utilized Matlab’s image processing toolbox to perform these operations, applying built-in functions for Gaussian filtering, edge detection (such as the Canny or Sobel method), and wavelet transforms (e.g., Haar wavelet) to achieve robust feature extraction.

  Feature encoding: Feature encoding is performed using vector quantization with the k-means clustering algorithm to encode the extracted pattern features. Vector quantization aims to reduce feature dimensionality, facilitating efficient storage and retrieval. This step’s mathematical formulation is expressed as follows^36,37: Given an eigenvector “x_i” and an encoding vector “s_i” of length “m”, the frequency “s_i(j)” with which eigenvector “x_i” maps to the “jth” cluster is represented by

  $${s}_{i}(j)=\mathop{\sum }\limits_{{x}_{{\rm{i}}}{\rm{\epsilon }}{C}_{j}}{{\rm{||}}{x}_{i}-{c}_{j}{\rm{||}}}^{2}$$

  (6)

  where “C_j” indicates the set of eigenvectors assigned to the “jth” cluster, and “c_j” represents the cluster center.

  Feature matching: Encoded eigenvectors of unidentified patterns are systematically compared to established reference eigenvectors within the database. The matching process involves quantitatively measuring the similarity or distance between encoded eigenvectors. By using Python-based scripts, the measured eigenvectors are effectively matched against stored reference data, enabling the reliable classification and recognition of Yi nationality patterns.

  The resulting eigenvectors encapsulate essential visual attributes, including texture, color features, contour shapes, grayscale information, directional lines, and proportional color-block distributions. This rigorous technical approach ensures the database’s robustness, supporting subsequent computerized pattern recognition and extraction tasks detailed in step 4.

• Step 4—Computerized pattern acquisition: Web crawlers are developed targeting a multitude of websites related to Yi nationality patterns, including those focused on cultural promotion and dissemination, shopping, clothing, historical relics, and cultural creativity. From these web pages, images are extracted and downloaded. The Python libraries “requests” and “Beautiful Soup” are utilized for retrieving web pages, parsing image links, and downloading the images. Once the image data is obtained, the pre-processing, eigenvector recognition, and saving procedures described in Step 2 are repeated.

• Step 5—Labeling of patterns: This step involves the collection, analysis, and annotation of traditional Yi patterns, drawing upon their esthetic characteristics, feature encoding, pattern interpretation, and common usage scenarios. A labeling system is then established, based on F (pattern), C (color), and S (composition), utilizing a one-to-many image-label relationship (PhotoID → label1, label2, label3). Each image is assigned a unique feature label, allowing designers to quickly access the label information for any given image.

• Step 6—Manual adjustment and improvement of database: Following the completion of the preceding five steps, a designer intervenes to perform final adjustments to the database. This includes verifying the accuracy and effectiveness of the data, assessing the feasibility of reintegrating previously removed images, and carrying out any other necessary operations. Moreover, the designer can enrich the database by generating original patterns through induction, abstraction, deformation, decomposition, and combination, thereby achieving pattern decomposition and transformation, optimizing and generalization, variation, and extension of patterns.

Construction of an image-generation model for traditional patterns

Following the curation and administration of the image database, the process proceeds to its crucial third phase: the development of the GAN model. In contrast to traditional handmade patterns, GAN models can rapidly produce numerous novel patterns by adjusting model parameters, thereby governing the patterns’ appearance and characteristics. This introduces fresh concepts and opportunities for the design and progress of traditional patterns. Deep learning image generation models, containing autoencoders, generative adversarial networks, and deep belief networks, etc, help image synthesis, information retrieval, and trans model conversions between images, text, and audio³⁸. This study will employ a GAN, considering its strong generative capabilities, unsupervised learning nature, capacity for diversity, scalability, and interpretability, to construct a generative model for traditional Yi patterns. This construction will unfold in three steps:

• Step 1—GAN model architecture: This study will utilize a deep convolutional neural network-based GAN (DCGAN) comprising a generator and a discriminator. The GAN seeks to create images, through the generator, sufficiently realistic to mislead the discriminator. The corresponding formulations for the generator and discriminator are presented in Eqs. (7) and (8). In these equations, “G” represents the generator, “D” denotes the discriminator, “z” indicates random noise, “θ“ expresses the parameters of the generator and discriminator, “x” depicts the input or generated pattern image, and H, W, and C, respectively, represent the height, width, and number of channels of the generated image.

  $$G({z;}\,{\theta }_{g}):z\,\epsilon \,{R}^{d}\to x\,\epsilon \,{R}^{H}\times^{{W}}\times^{C}$$

  (7)
  $$D({x;}\,{\theta }\_{d}):x\,\epsilon \,{R}^{H}\times^{W}\times^{C}\to [0,1]$$

  (8)

  The objective function of the GAN model includes the loss functions of both the generator and discriminator, and measures the difference between generated and real images³⁹. The mathematical representations of these loss functions are given in formulas (9) and (10). In these formulas, “L” represents the loss function, “z” denotes random noise, “pz(z)” expresses the noise distribution, “PDATA(x)” indicates the real image distribution, “x” depicts a real pattern image, and “1−D(G(z))” illustrates the discriminator’s discrimination result on the generated image.

  $${L}_{{\rm {G}}}=-\frac {1}{2}\,E_{\{{z \sim {{p}_{z(z)}\}}\left[\log(D(G(z)))\right]}}$$

  (9)
  $${L}_{{\rm {D}}}=-\frac {1}{2}\,E_{\{{x \sim {p}_{{{data}}(x)}\}\,[\log (D(x))]}}-\frac {1}{2}E_{{\{{z \sim {p}_{z}(z)\}}[\log (1-D(G(z)))]}}$$

  (10)

• Step 2—GAN model training: The generator and discriminator are trained independently and iteratively, with the loss function monitored until the GAN model’s loss function converges⁴⁰. First, the discriminator is trained on real images, its parameters are updated through gradient descent to reduce its loss function. Then, the generator is trained to produce images from noise, which are then passed to the discriminator. This process also decreases the loss function through gradient descent applied to the generator’s parameters^41,42,43,44. The corresponding mathematical representations are detailed in formulas (11) and (12). In these formulas, “θ“ represents the parameters of the discriminator or generator, “α“ denotes the learning rate, and “∇“ expresses the gradient of the discriminator’s or generator’s loss function with respect to the discriminator’s parameters. This iterative parameter update through gradient descent derives from the network’s ability to create and recognize images.

  $${\theta }_{{{d}}}={\theta }_{{{d}}}-\alpha {\nabla }_{{\{{\theta }_{{{d}}}\}}{L}_{{\rm {D}}}}$$

  (11)
  $${\theta }_{{{g}}}={\theta }_{{{g}}}-\alpha {\nabla }_{{\{{\theta }_{{\rm {g}}}\}}{L}_{{\rm {D}}}}$$

  (12)

• Stage 3—Pattern image synthesis: Following the training of the GAN model with the complete codebase, the generator component can be employed to synthesize novel pattern images. By introducing random noise as input, the generator produces new pattern images, thereby achieving the innovative generation of traditional Yi patterns.

  In summary, these three stages enable the construction of an image synthesis model for traditional Yi patterns. Simultaneously, the innovative application of generative models offers new avenues and concepts for the innovation and development of traditional patterns.

Application and evaluation of innovative patterns

After the successful construction and generation of pattern databases, image models, and simple patterns, an evaluation of the synthesized images is necessary before their practical application. This study adopts three methods: DR, MOS, and FID, to evaluate the generated pattern images. DR measures image diversity and uniformity by calculating inter-cluster distances. MOS is a subjective evaluation method based on human ratings, while FID compares the distributions of real and generated images³⁸. A lower FID value indicates higher quality. The procedures for these three methods are detailed in formulas (13), (14), and (15).

$${{{DR}}}=\sum (1-D(i,j))/(k(k-1)/2)$$

(13)

$${{{MOS}}}=(\sum {qi})/n$$

(14)

$${{FID}}=\left\|\mu^{1}-\mu^{2}\right\|^{2}+{{Tr}}\left({\Sigma}^{1}+{\Sigma}^{2}-{2}({\Sigma}^{1}{\Sigma}^{2})^{\{\frac{1}{2}\}}\right)$$

(15)

In formula (13), “D” represents the distance between clusters, “k” denotes the number of clusters, and the value range of DR is [0,1]. If the score is higher, the image will be more diverse. In formula (14), “qi” expresses the score of the “ith” subject, “n” indicates the number of subjects, and the range of MOS values is^1,5. If the score is higher, the image quality will be higher. In formula (15), “μ¹” and “μ²”, respectively depict the mean vectors of the real image and the generated image, “∑” explains the covariance matrix of the image, “Tr” conveys the trace operation of the matrix, and the range of FID values is [0, ∞). If the score is smaller, the image quality will be better.

These three evaluation indicators can help designers evaluate the diversity, realism, and visual quality of generated images, and determine the advantages and disadvantages of the generated model. This can be applied in various physical fields for creative design and expression and realize the diversified application of innovative patterns in social life. The specific form and physical carrier are demonstrated in supplementary information.

Results

Construction of Yi traditional pattern database

Having established the methods for establishing the database and constructing the GAN model, this study will begin to manually collect traditional Yi nationality patterns and analyze relevant esthetic features to preliminary construct the database.

The artistic traditions of the Yi nationality, including their patterns, have evolved over millennia, resulting in established patterns, color palettes, and compositional structures. These patterns, born from creative interpretation and optimization of lived experience, exhibit both standardization and broad applicability once established. In addition to their purely visual characteristics, Yi patterns embody the social and cultural values of the community. A foundational database of these traditional patterns has been compiled through the collection, analysis, and annotation of examples, focusing on feature encoding, pattern interpretation, and typical usage. A portion of this database is presented in Table 1.

The labeling system outlined in Table 2 is structured around three core elements: pattern, color, composition, and form. Yi nationality compound patterns arise from the combination of basic patterns in two primary aspects. First, basic patterns with different meanings can be joined to create new, meaningful designs. Second, regular patterns can be produced by arranging and combining established elements in structured ways. The main colors in Yi patterns are black, yellow, and red. Black, the primary ground color, signifies power and prestige. Yellow represents the sun, moon, and earth, reflecting a reverence for nature and a wish for abundant harvests and prosperity. Red, symbolizing fire, embodies courage and passion. Among the Yi people of Liangshan, common compositional forms include repeating patterns (such as bipartite continuous patterns, continuous patterns), symmetrical patterns (such as center divergent, two-sided symmetrical, four-sided symmetrical, etc.), isolated patterns, corner elements, and scattered placements. Composite patterns integrate these forms to create rich formal effects, conveying rhythm and visual cadence in a clear hierarchical structure.

Table 2 Specific forms and physical carriers of diversified applications of new patterns

Categorizing the visual information resources of Yi traditional patterns according to the three classes of F (pattern), C (color), and S (composition form) facilitates the development of a bidirectionally structured database for encoding image features. Each category in the database contains several classification tags, enabling users to retrieve the necessary information through tag selection. Multiple tags can be selected in each category. For instance, the digital sample with ID P4 corresponds to multiple feature labels (P4 → F3 wavy pattern, F4 golden chain pattern, F5 rapeseed pattern, C3 black, yellow, red, S3 center divergence). Designers can readily retrieve information based on these labels. Searching with feature label F1 produces samples P1 and P4 in Table 1. A search employing the combined criteria of F1 + S1 obtains only sample P1.

It is essential to then obtain a large number of Yi pattern images and, based on this primary database, construct a pattern database that is as complete as possible. Throughout this data collection process, 6478 patterns were collected manually, resulting in 2478 valid pattern images after processing. An additional 50,000 patterns were collected computationally, deriving 5973 valid pattern images after processing. Following pre-processing, screening, collation, testing, encoding, and further optimizations, a total of 8451 valid pattern data images were archived in the database.

Finally, a computer-based digital application platform for Yi nationality patterns was developed from the database content. This platform facilitates the selection, extraction, display, and retrieval of pattern images. The interface, depicted in Fig. 3, enables users to readily access the visual representation and identification number of each pattern. In addition, users can appraise the feature codes, descriptions, and usage examples of patterns through clicks and navigation on the platform. This digital platform offers enhanced efficiency in the storage, preservation, and transformation of Yi nationality patterns. It constitutes a real-time updatable database represented by multi-level tagging, a well-defined classification system, and superior image resolution.

Fig. 3

Digital application platform for traditional patterns of Yi nationality.

Construction of an image-generation model for traditional patterns of Yi nationality

Then, this research develops an image generation model for traditional Yi patterns. In the generator, the input noise vector has a dimensionality of 100. The generator architecture incorporates four transposed convolutional layers, each with a kernel size of 4*4 and a stride of 2, where the number of channels progressively decreases. The complete process maps the noise vector to a higher-dimensional feature space through a fully connected layer, then gradually reduces the dimensionality of the feature space utilizing transposed convolutional layers. Finally, it outputs an image matching the dimensions of the original image. The discriminator also employs four transposed convolutional layers, maintaining consistent kernel size and stride while progressively increasing the number of channels. The discriminator’s process involves inputting pattern images, extracting image features utilizing a series of convolutional layers, and finally mapping these features to a scalar value through a fully connected layer. This value signifies the probability of the input image being a genuine image. Simultaneously, a sigmoid activation function maps the output to a range of 0–1, representing the output of the discriminator.

During training, the generator and discriminator parameters were updated utilizing the Adam optimizer with a learning rate of 0.0002. Dropout, with a probability of 0.5, was applied to the convolutional layers of the discriminator and the transposed convolutional layers of the generator to prevent overfitting. LeakyReLU activation functions were employed in both the generator and discriminator to expedite convergence. As illustrated in Fig. 4, the loss function decreased as the number of training iterations increased. Figure 4 plots the loss function (vertical axis) against the number of training iterations (horizontal axis). Experimental findings indicate a correlation between pattern complexity and the required number of training iterations: more complex patterns, exhibiting more obvious stylistic features, proved more challenging in terms of feature extraction, thus necessitating more iterations. The traditional pattern analyzed in this study achieved satisfactory results after approximately 850 training iterations.

Fig. 4

The training process of generative adversarial networks.

While the generated images from the above experimental processes successfully integrated pattern features, certain stylistic aspects, notably structural proportions and color composition, demonstrated instability. To address this, color attribute loss and structural similarity indicators were added to the GAN model for traditional Yi patterns. This change facilitates the generator in better learning the color and structural features of Yi patterns, finally producing more realistic and typical pattern images. The specific steps are detailed below:

• The first step defines the objective functions for both the generator and discriminator. A new objective function is derived by adding color attribute loss and structural similarity indicators to the original GAN objective function.

• The second step computes the color attribute loss. This loss is calculated for each training batch by comparing the generated images against the target pattern.

• The third step involves calculating the structural similarity index. For each training batch, the generated images and target patterns are converted to grayscale, and their structural similarity index is computed. Then, the parameters of both the generator and discriminator are updated. Gradients for the generator and discriminator are calculated utilizing backpropagation, and the parameters are updated through the Adam optimization algorithm⁴⁵. Formulas (16)–(18) detail the computation of the new objective function, the color attribute loss value, and the structural similarity index, respectively.

$$\mathop{\min }\nolimits_{{{G}}}\mathop{\max}\nolimits_{{{D}}}V(D,G)+{\lambda }^{1}{{{{Loss}}}}_{{{{color}}}}+{\lambda }^{2}(1-{{{SSIM}}}(G(z),Y))$$

(16)

$${{{{Loss}}}}\_{{{{color}}}}=\mathop{\sum }_{i=1}^{N}{(G{({z})}_{i}^{{\rm {c}}}-{Y}_{i}^{{\rm {c}}})}^{2}/N$$

(17)

$${{{SSIM}}}(G(z),Y)=((2{\mu}\_\{G({\rm{z}})\}{\mu}\_{Y}+{C}_{1})(2{\sigma }\_\{{{\rm{G}}({\rm{z}}),{\rm{Y}}}\}+{C}_{2}))/(({\mu }\_{\{G(z)\}}^{2}+\,{\mu }\_{Y}^{2}+{C}_{1})({\sigma }\_{\{G(z)\}}^{2}+{\sigma }\_{Y}^{2}+{C}_{2}))$$

(18)

In formula (16), “V(D, G)” represents the original objective function, “λ₁” and “λ₂” denote the weights of color attribute loss and structural similarity index, respectively. In formula (17), “Y” indicates the color attribute of the target pattern, “N” expresses the total number of pixels, and “G(z)^c” and “Y^c” depict the color values of the generated image and the target pattern at a certain point in the pixel. In formula (18), “μ” conveys the generated image mean, “σ” illustrates the standard deviation, C₁ and C₂ are constants.

This iterative process continues until the generator produces a pattern image that satisfies the predefined color characteristics and structural similarity criteria, resulting in the automatic creation of pattern images with Yi nationality characteristics. Example generations and their respective processes are illustrated in Fig. 5. As presented in Fig. 5, the resulting pattern structures are well-defined, exhibiting clean lines and ready recognizability. The overall esthetic and color palettes align with the characteristics of traditional Yi patterns. The color schemes primarily utilize black, yellow, red, and blue. The compositional arrangements maintain traditional pattern characteristics, such as continuity, symmetry, and dispersion. Regarding pattern morphology, these generated designs originate from abstraction, deformation, decomposition, combination, change, and extension of fundamental units, including wave, golden chain, and rapeseed patterns.

Fig. 5

The generation process of some innovative patterns.

The stability and innovation demonstrated across these three facets contribute to a stronger sense of personalized artistry in the patterns. However, as most patterns in the source database are hand-crafted, their lines often deviate from perfect straightness. Therefore, generated patterns can similarly present with curved lines and irregular borders. Prior to evaluation and implementation, these novel patterns require further image optimization utilizing computer software to enable later image quality evaluation and diverse applications.

Image quality evaluation results of Yi nationality innovative patterns

Following the development of the Yi nationality pattern database and image generation model, a comprehensive and objective evaluation of image quality is undertaken to determine the suitability of these patterns for adaptation and integration across various design contexts. First, 100 novel patterns were generated at random utilizing the image generation model and received minor adjustments utilizing Adobe Photoshop. Excessively curved lines or complicated borders were simply processed to enhance their applicability following evaluation, taking care to preserve color, shape, linework, and other qualities. A selection of the patterns employed in the image quality evaluation is presented in Fig. 6.

Fig. 6

Partial patterns for image quality evaluation.

Then, image quality was evaluated by DR, MOS, and DIF methods, with the results presented in Table 3. The DR assessment indicates that the Yi traditional patterns produced by GAN attained high scores for diversity and stability; however, there remains a certain degree of variation or instability. The median and mode values of 0.84 and 0.85, respectively, suggest that most images demonstrate similar diversity and stability. The low variance and standard deviation point to relatively consistent scores for the data-integrated images. For the MOS assessment, ten experts and scholars specializing in Yi culture offered scores. Results demonstrate that the GAN-generated traditional Yi patterns exhibited strong visual appeal, yet discrepancies were observed. With a median of 4.5 and a mode of 4.6, most images presented similar visual quality. The high standard deviation, however, indicates differences in evaluator perceptions of image quality. FID assessment results exhibit that the feature distribution of GAN-generated traditional Yi patterns closely approximates that of authentic data, albeit with certain deviations. A mode of 28.3 suggests that the feature distribution of certain images is relatively similar to the real data. Nevertheless, both the standard deviation and variance are relatively large, indicating significant differences in the evaluation results.

Table 3 Image quality evaluation results of Yi nationality innovative patterns

In summary, these data demonstrate the GAN-generated Yi traditional patterns’ strong performance in diversity, stability, and visual quality. While the feature distribution is average and there are certain differences and instability in the evaluation data, the overall quality of the patterns is high. This meets the requirements for diversified design applications in the future (Table 4).

Table 4 The characteristics of innovative patterns of Yi nationality

Discussion

This study demonstrates that the novel Yi patterns generated by generative adversarial networks (GANs) can be applied in diverse fields such as apparel, home decor, architecture, and art, contributing to the preservation and promotion of Yi culture while offering potential commercial value. However, further research is required to assess the actual impact of these patterns in real-world commercial applications.

From the 100 generated images, the ten highest-ranked ones were selected based on overall quality, with a focus on colors such as black, yellow, red, and blue. Four of these were chosen as design inspiration. These patterns were analyzed according to Yi esthetic principles, with their composition, color palette, and symbolic meaning cataloged and theorized for potential applications. For instance, one pattern symbolizing harvest and hope could be applied in Yi clothing design, while another, symbolizing divine protection, could be used for Yi jewelry and packaging.

The GAN-generated textures offer creative inspiration for designers, though their application in real-world designs requires further refinement to meet specific needs. In supplementary information, four design teams used these patterns to create derivative designs in scarf, packaging, paintbrush, and clock designs, drawing on visual communication and product design methodologies. These designs successfully integrated Yi patterns into modern applications, showcasing their cultural significance.

This work introduces a GAN-based framework that combines esthetic analysis and generative modeling, distinguishing it from previous static preservation methods. While our approach facilitates the preservation and reinterpretation of Yi patterns, its limitations include the quality of the generated patterns being influenced by the database size, and the subjective evaluation methods employed. Expanding the database, incorporating more automated evaluation techniques, and validating the practical application of this method in commercial contexts will be key areas for future work.

The present study contributes a novel methodology for reconstructing traditional ethnic patterns using a GAN-based model, aiming at cultural preservation and innovation. The method comprises four steps: esthetic analysis, database construction, GAN model development, and application assessment. These steps enable systematic protection and creative reinterpretation of traditional Yi patterns. Specifically, the phases of esthetic analysis and database construction ensure cultural authenticity and richness, while GAN modeling and application assessment facilitate the innovative application and commercialization of these patterns.

The major contributions of this study include:

• Promoting sustainable development of Yi cultural heritage by introducing diverse cultural elements into contemporary designs;

• Enhancing design efficiency by reducing costly and labor-intensive image acquisition procedures, thus accelerating creative processes;

• Expanding the practical applications and commercial possibilities of Yi patterns, leveraging their unique esthetic appeal.

However, we acknowledge limitations, notably the dependence on database size and subjective evaluation methods. Future research should focus on database expansion, automated pattern evaluation, and verifying practical applicability in commercial scenarios. Despite these limitations, this study represents a significant step toward integrating traditional culture with AI-driven innovations.