Chinese Segmentation Research Articles

Stroke‐Seg: A deep learning‐based framework for chinese stroke segmentation

AbstractChinese stroke segmentation is a crucial and challenging task for various downstream applications such as font generation, aesthetic evaluation etc. Conventional semantic segmentation techniques typically face difficulties in accurately segmenting strokes, as intersection regions in Chinese characters can belong to multiple strokes simultaneously, and these approaches often lack a holistic understanding of character composition. This paper proposes a character stroke segmentation framework named Stroke‐Seg that integrates with various semantic segmentation architectures, demonstrating adaptability to different backbone networks to tackle the above tasks. A multi‐label output strategy is proposed to effectively classify strokes in intersection areas, overcoming the limitations of traditional semantic segmentation approaches. Additionally, a prior knowledge vector is incorporated into the input layer to provide character‐specific information on stroke composition, enhancing the ability of the framework to precisely identify and segment strokes. The effectiveness of the proposed framework is demonstrated through evaluations of a comprehensive dataset (brush calligraphy stroke segmentation dataset). Evaluations show that the proposed framework significantly improves the capability of semantic segmentation networks, achieving a remarkable improvement of up to 19.9% in the true stroke rate, compared to traditional stroke segmentation techniques. The Stroke‐Seg framework integrated with TransUNet demonstrates its high performance with an impressive 98.2% true stroke rate. Furthermore, the proposed framework combined with FCN still achieves good performance while consuming the least amount of computational resources and memory, demonstrating its potential for lightweight design.

Fine Segmentation of Chinese Character Strokes Based on Coordinate Awareness and Enhanced BiFPN.

Considering the complex structure of Chinese characters, particularly the connections and intersections between strokes, there are challenges in low accuracy of Chinese character stroke extraction and recognition, as well as unclear segmentation. This study builds upon the YOLOv8n-seg model to propose the YOLOv8n-seg-CAA-BiFPN Chinese character stroke fine segmentation model. The proposed Coordinate-Aware Attention mechanism (CAA) divides the backbone network input feature map into four parts, applying different weights for horizontal, vertical, and channel attention to compute and fuse key information, thus capturing the contextual regularity of closely arranged stroke positions. The network's neck integrates an enhanced weighted bi-directional feature pyramid network (BiFPN), enhancing the fusion effect for features of strokes of various sizes. The Shape-IoU loss function is adopted in place of the traditional CIoU loss function, focusing on the shape and scale of stroke bounding boxes to optimize the bounding box regression process. Finally, the Grad-CAM++ technique is used to generate heatmaps of segmentation predictions, facilitating the visualization of effective features and a deeper understanding of the model's focus areas. Trained and tested on the public Chinese character stroke datasets CCSE-Kai and CCSE-HW, the model achieves an average accuracy of 84.71%, an average recall rate of 83.65%, and a mean average precision of 80.11%. Compared to the original YOLOv8n-seg and existing mainstream segmentation models like SegFormer, BiSeNetV2, and Mask R-CNN, the average accuracy improved by 3.50%, 4.35%, 10.56%, and 22.05%, respectively; the average recall rates improved by 4.42%, 9.32%, 15.64%, and 24.92%, respectively; and the mean average precision improved by 3.11%, 4.15%, 8.02%, and 19.33%, respectively. The results demonstrate that the YOLOv8n-seg-CAA-BiFPN network can accurately achieve Chinese character stroke segmentation.

The application of Hyperpixel segmentation algorithm in Chinese painting

Chinese painting, as an important component of traditional Chinese art, not only expresses Chinese culture and history, but also contains rich artistic connotations. Therefore, extracting and analyzing the main objects of Chinese painting is of great significance. The research introduces a fusion deep network based super pixel segmentation algorithm for image segmentation, and utilizes user marked segmentation target objects and backgrounds. The maximum similarity region merging idea is adopted to extract the main subject objects in Chinese painting, thereby achieving a deep understanding of the work. The results show that the boundary recall and segmentation accuracy of the Chinese painting subject object segmentation algorithm based on super pixels on the BuptPainting dataset are as high as 94.38% and 98.06%, respectively, with an undersegmentation error rate of only 7.69%. Meanwhile, the average accuracy value, optimal data scale value, and optimal image scale value of this method on the BuptPainting dataset are as high as 0.657, 0.541, and 0.529, respectively. The method proposed by the research institute has significant advantages in segmenting the main objects of Chinese painting, providing new ideas and methods for image analysis and processing of modern Chinese painting, and helping to further explore the artistic charm and cultural connotations of Chinese painting.

An approach for handwritten Chinese text recognition unifying character segmentation and recognition

Text line recognition methods can be categorized into explicit segmentation based and implicit segmentation based ones. Explicit segmentation based methods require character-level annotation during training, while implicit segmentation based methods, trained on line-level annotated data, face alignment drift challenges. Though some methods have been proposed to address these challenges using weakly supervised object detection, they often rely on cumbersome pseudo-box generation processes and complex decoding. In this paper, we propose a unified framework to overcome these challenges, achieving high accuracy in text recognition and character segmentation. To eliminate the need of character-level annotated real text line data in training, we introduce a novel training paradigm that utilizes character-level annotated synthetic data and line-level annotated real data jointly. For synthetic data, candidate characters are explicitly aligned with labeled characters to generate hard labels for supervising model training. For real data, implicit alignment is produced by Connectionist Temporal Classification (CTC) mapping to provide soft labels for weakly-supervised model training. And for inference, we propose two decoding strategies leveraging the advantages of Non-Maximum Suppression (NMS) and CTC decoding. Extensive experiments on benchmark datasets demonstrate the superior performance of our method in text recognition and character localization, even with minimal amounts of character-level annotated line data.

An Intelligent Medical Q&A System Based on Natural Language Processing

Aim: To improve the accuracy of Chinese word splitting. Background: With the development of Internet technology, people want to get some effective medical information from the Internet, but there are still technical difficulties for non-specialists. At the same time, the level of medical construction can not keep up with the demand of patients for medical treatment, the phenomenon of doctor-patient conflicts has not been fundamentally solved, and the problem of difficult consultation prevails. With the arrival of the era of big data and artificial intelligence, medical Q&A has been applied. Objective: In order to meet the user's need to get the correct answer as soon as possible, medical Q&A needs to have high execution efficiency. The accuracy of Chinese participle directly affects the execution efficiency of Q&A. Improving the accuracy of Chinese participle can fundamentally improve the accuracy of medical Q&A and shorten the answering time. Methods: Improvement of the Chinese Segmentation Algorithm based on BI-LSTM-CRF using natural language processing technology. Based on the same medical Q&A dataset, the medical Q&A is trained and tested under three commonly used segmentation algorithms and the segmentation algorithm designed in this paper. Results: The experiments show that the Chinese Segmentation Algorithm studied in this paper improves the accuracy of medical Q&A and can improve the execution efficiency of medical Q&A. Conclusion: Based on the calculation and matching process of the same similar answers, different word-splitting methods directly affect the effect of medical Q&A in the later stage. The better the effect of segmentation, the higher the accuracy of the correct answers in medical Q&A. The improved LSTM-CRF split word accuracy designed in this paper achieves a good split word effect in the training process. Compared with the HMM segmentation algorithm, which has the best segmentation performance among the other three algorithms, the segmentation accuracy is improved, and the accuracy of the Q&A that delivers the correct answers is relatively high. Despite the improved accuracy in segmenting the medical dataset, the time complexity did not decrease much. The LSTMCRF combined network segmentation algorithm designed in this paper performs better in medical Q&A compared to other commonly used segmentation algorithms in terms of subject operating characteristics and larger regions surrounded by coordinate axes.

A hybrid method of combination probability and machine learning for Chinese geological text segmentation

To address the issues surrounding incomplete coverage of core dictionaries, limited training corpora, and low precision in Chinese geological text segmentation, a knowledge- and data-driven word segmentation method by combining combination probability and machine learning was proposed in this paper. We extracted mathematical feature information from terms in Chinese geological text to construct a Term Combination Probability Model (TCPM) for Chinese word combinations by integrating the combination features of geological terms and the Chinese writing styles under zero-sample conditions. The TCPM was used to extract geological terms with high combination characteristics as a user-defined dictionary, and then a geological corpus was constructed by using a general domain word segmentation method based on this dictionary. After a small amount of manual review and optimization, the geological corpus was trained with a BiLSTM-CRF model to segment Chinese geological text. The proposed method in this paper was tested using a regional geological survey report set in Henan Province, and the precision, recall, and F1-score of the method are 92.65%, 92.53%, and 92.59%, respectively. The experimental results demonstrated that this method, combined with the inherent knowledge features of geological text and the machine learning method, can assist in expanding the core dictionary for Chinese geological text segmentation based on zero-sample, and can improve the segmentation precision of Chinese geological text compared to simple general word segmentation methods or machine learning methods.

On Some Possible Differences between Punctuation Marks in the Chinese and Spanish Languages

Languages vary in spelling and grammar. Starting from Didactics in the field of Language and Applied Linguistics, our research is conducted on suprasegmental features: contrastive analysis for future Spanish teachers to native Mandarin speakers. Punctuation marks are essential for correctly understanding any written text in any language. However, they are not the same for each language, nor do they have the same usage or form. This work formulates a research proposal to compare the following punctuation marks in Chinese and Spanish: period or full stop, comma, question marks, and hyphen, which are the most commonly used and, above all, because the differences between them are more representative. To achieve this objective, it starts from the deficiencies in the use of these marks in both languages and reflection on the historical ignorance regarding their usage, form, and possible similarities between Chinese and Spanish segments. We conclude with the idea that the formal dissimilarity between certain Spanish and Chinese punctuation marks, such as the full stop and the question mark, does not cause extraordinary difficulties for Chinese speakers in acquiring their correct use in Spanish. Those signs whose forms share similarities, such as the comma and quotation marks, do represent acquisition difficulties. In order to specify the scope of this phenomenon on punctuation, in the future it will logically be necessary to study more punctuation signs, such as semicolons, colons, as well as ellipses.

A Study on Chinese Consumer Market Segmentation by DMA: Focusing on Consumer Behavior and Choice Attributes for Korean Automobiles

A Study on Chinese Consumer Market Segmentation by DMA: Focusing on Consumer Behavior and Choice Attributes for Korean Automobiles

Discovery, geological setting and mineralization of the Shuvuutai molybdenum deposit, central-southern Mongolia

We document the newly discovered Shuvuutai Molybdenum (Mo) porphyry deposit located in the central-east Mongolian volcanic/-plutonic belt of the southern Altaids. The deposit is characterized by quartz vein- and veinlet-disseminated-type Mo ore bodies mainly hosted in a Late Permian-Early Triassic granitic intrusion. In this study, we provide a detailed study that integrates detailed geology with petrographic, mineralogic, geochemical, and geochronologic date to determine the tectonic setting of the deposit. Fluid inclusion analysis is used to reconstruct the fluid evolution history of the Shuvuutai Mo mineralization system and to understand the origin of the deposit. The ore bodies occur mainly as veins, lenses, and pods in positions from inner intrusions through contact zones to the host rocks distal to causative intrusions. The host rocks are variable in lithology, including granites, porphyries, volcanic breccias and tuffs, and sedimentary rocks. Outward from ore bodies to host rocks, the wall rock alteration is zoned from potassic (K-feldspar-quartz-mica), through phyllic (quartz-sericite-chlorite-epidote), to propylitic or argillic alterations. Two samples from porphyry stock intrusions at Shuvuutai were selected for U–Pb zircon geochronology and trace element composition studies, including the ore-hosting monzodiorite porphyry and quartz syenite porphyry. Our new ages are consistent with, or slightly postdate published zircon U-Pb ages of early phase of monzodiorite porphyry (268–232 Ma) and the main phase of quartz syenite porphyry (251.6 ± 1.7 Ma). Hydrothermal mineralization generally includes four stages, from early to late, represented by (1) potassic feldspar-quartz veins or veinlets, (2) quartz-molybdenite stockworks, (3) quartz-sulfide stockworks, and (4) quartz ± carbonate veins or veinlets. The ore-forming fluids were initially magmatic in origin and show high-temperature and high-salinity, containing daughter mineral and CO2-bearing fluid inclusions; and eventually evolved to low-temperature, low-pressure, low-salinity and CO2-poor meteoric water. S isotopic compositions of sulfides from the Shuvuutai Mo porphyry deposit indicate that the source of the ore-forming material was derived from the felsic magma. The Central-Eastern Mongolian volcanic-plutonic belt continues from Mongolia into northeast China, where similar Mo deposits are not yet recognized, suggesting that a new exploration strategy, based on the characteristics and history of the Shuvuutai discovery, is needed for the Chinese segment of the Altaids.

A Novel Chinese Address Segmentation Method with Self-growth Feature

Chinese Address Segmentation (CAS) is a crucial step that can greatly enhance the performance, accuracy, and reliability of geo-coding technology. However, it presents a tremendous challenge due to the inherent lack of obvious word boundaries, complex grammatical and semantic features. To address this challenge, we propose a novel CAS method or model that starts from scratch, without relying on any pre-installed knowledge about Chinese addresses. Instead, it dynamically evolves and grows its knowledge library by leveraging contextual information and comparing addresses during the process of dividing them into address elements. Our approach does not rely on Chinese language or address-element dictionaries, nor does it depend on address statistics. The knowledge library is automatically extracted and organized in a tree data structure. This unique approach allows our method to effectively segment addresses from any area of China, including regions with intricate address expressions, such as the Inner Mongolia Autonomous Region. Experimental results demonstrate that our method achieves high precision in address segmentation.

ОТВЕТСТВЕННОСТЬ РАБОТНИКОВ КАК ОБЪЕКТ УПРАВЛЕНИЯ: КОМПАРАТИВНЫЙ АНАЛИЗ КИТАЙСКИХ И РОССИЙСКИХ ПОДХОДОВ

The paper presents the results of a desk study of the responsibility at work, the purpose of which is to determine the degree of relevance of personnel responsibility issues, to identify the prevailing perception stereotypes of the responsibility by Chinese and Russian employees. The collection and processing of empirical data was carried out using the methods of quantitative and qualitative content analysis of text arrays in the Internet Chinese and Russian segments. An online survey of Russian enterprises’ employees involved in the poll using the spontaneous sampling method allowed the authors to identify the respondents attitude to the problem under study, as well as comparison of the Chinese and Russian employees’ narratives analysis results has practical importance for finding effective methods and tools for educating responsible employees, developed on the best practices basis. One of the key conclusions based on the study results is the disproportion between the degree of Russian specialists’ attention to the issues of personnel responsibility management and the significance of the influence of the employee responsibility factor on the organization’s performance indicators.

한국 화장품의 대중국 수출 촉진 요인과 유통채널 설계방안에 관한 연구

The Chinese cosmetics market is showing rapid growth. This study analyzed causal relationship between Korean cosmetics exports and China’s macroeconomic facilitation factors. After analyzing Chinese cosmetics market and challenges for Korean cosmetics companies, this paper suggests an effective distribution channel selection and management strategies for Korean cosmetics companies. Indirect distribution, optimum route plan for small and medium-sized cities from metropolises, utilization of electronic commerce, differentiation between premium brands and diffusion brands and strategic quanxi orientation shall be required for most of the Korean cosmetics companies.
 The limitations of this study are as follows: analysis of export promotion factors of Korean cosmetics export in Chinese market are limited to a few variables and that this paper does not cover Chinese regional market segment’s characteristics. However, this paper has significant implications for understanding on the distribution channel strategies for Korean cosmetics export in China and suggests some strategic solutions.

Automatic sentence segmentation for classical Chinese: The Spring and Autumn Annals as an example

Abstract There exists no sentence boundary in most classical Chinese literature texts. Since it is difficult to read literature of this kind, experts in literature or linguistics would segment the sentence manually. This article explores the effectiveness of classical Chinese sentence segmentation method so as to provide a reference for classical Chinese punctuation. On the basis of the machine learning methods, we chose three components of machine learning, namely models, tagging schemes, and features, to compare the learning results. The models include conditional random field (CRF) models, long short term memory (LSTM) models, BiLSTM–CRF models, and three Bidirectional Encoder Representation from Transformers (BERT) models. There are five tagging schemes in this article and three features including the statistical feature, Guangyun, and Fanqie. Finally, the performance of the combined feature template is evaluated by ten-fold cross-validation on four classical Chinese texts in different genres. The SikuBERT model is proved to be the most effective model for sentence segmentation at present. Different tagging schemes and various features are introduced. The results show that 5-tag-J tagging schemes can improve performance. Statistical feature, as an important clue for classical Chinese sentence segmentation, is useful in related tasks, but Guangyun and Fanqie have little impact. Other important factors of sentence segmentation are genres and writing styles.

The Influence of the Rise of Chinese Feminism on the Marketing Strategy of Beauty Products

This research study is based on the realistic background of the rise of feminism in China, which noticeably influenced female consumers, specifically through the spread of the internet. Due to the high activation of the topic of feminism, this study plans to discover its influence on the beauty market in China, as female consumers occupy a relatively higher proportion in this market. This research has a deep background of the development of the Chinese beauty industry since reform and opening up in China. The marketing principles in the Chinese beauty market are then discussed and analyzed while supported by a questionnaire based on the "snowball" sampling method. It leads to Chinese consumer segmentation based on consumers' definitions of feminism. Throughout the research, consumers' willingness to see feminist elements in beauty advertisements and the variety of feminism that consumers have defined have been discovered. It hints at the future patterns of "how to market" to beauty organizations in the Chinese market. This research study can help widen the range of research in the academic marketing field and provide more ideas to academicians in future marketing coupled with a way to think from a new angle. It could also possibly help business people who work in the beauty industry in the Chinese market to have an idea of how to present consumers and potential consumers think and how to manipulate and reflect under the condition of fast development on feminism.

Design and Implementation of a Recommendation System for Buying Fresh Foods Online Based on Web Crawling

As shopping patterns have gradually shifted from offline to online mode, and with recent lockdowns during the coronavirus disease 2019 (COVID-19) pandemic restricting foreign trade and accelerating the growth of the domestic economy, digital transformation has become a major strategy for many retailers to support and expand their businesses. With the pandemic becoming a turning point, the business of major e-commerce companies in Taiwan in the retail of dry goods has grown significantly, and it has driven the online sales of fresh products as well. In this era of fierce competition, it is especially important to find a way that enables consumers to quickly find ideal fresh products on multiple platforms, shortens the time for price comparison, and improves the efficiency of online shopping. This study uses the Python programming language to write a web crawler program that captures product information from fresh food e-commerce platforms, including product introduction, price, origin, and sales volume, and then defines the relevant status of the product, such as product popularity. Accordingly, through Chinese text segmentation and term-frequency calculation, it aims to classify the product names and introductions into frequently occurring words and use them as product-related labels. Finally, the program combines the product information processing results and product-related labels to construct an online fresh food recommendation system. The results of the proposed system show that it reduces the time and energy spent comparing prices. It can also guide consumers to browse products that may be of interest using relevant tags and increase consumption efficiency by helping them find the ideal item when shopping.

Virtual scene generation promotes shadow puppet art conservation

AbstractAs an ancient performing art, shadow puppetry is a treasure of Chinese art. However, with the development of society, shadow puppetry becomes less well‐known among the young generation. To preserve and further spread this traditional culture, the digitization of shadow puppets is playing an increasingly important role in shadow puppetry conservation. Despite this, the spread of shadow puppetry culture in modern times still faces many hindrances. In digitalized shadow puppet art, the virtual scenes in certain degree determine the artistic effect of shadow puppet performance. The commonly used method for digital shadow puppet scene construction is via artificially created models which are then placed in corresponding positions. Obviously, this is a cumbersome and time‐consuming task. Therefore, a semantic‐based scene generation method for digital shadow puppet performance scene is proposed in this paper. According to this method, the key information is extracted from the descriptive text using the Chinese text segmentation technology. Meanwhile, we generate semantic scene graphs and search the corresponding shadow puppet models in the model library to construct the virtual scenes of digital shadow puppet performance. In the evaluation experiment, we invited 30 volunteers (10 female and 20 male) who had been exposed to traditional shadow puppet play in their daily lives. As suggested by the experimental results, the digital shadow puppet performance scene generated in this paper exhibit advantages of convenient use and high availability, which largely enhance the effect of digital shadow puppet performance. It should nonetheless be noted that it's not easy to extract the spatial relations from complicated texts, which inevitably limits the effectiveness of our scene generation method. The research work of this paper aims to promote digital shadow puppet technology and provide insights for the inheritance and conservation of traditional shadow puppet art.

Price comovement and market segmentation of Chinese A- and H-shares: Evidence from a panel latent-factor model

This paper examines the price comovement of cross-listed Chinese A- and H-shares using a panel model with latent factors and a heterogeneous long-run structure. Our model is more flexible than the cointegration system and is estimated using the data-driven Cup–Lasso method. The long-run H-share price discounts are heterogeneous across different groups of stocks. We have identified both stationary and nonstationary latent factors in the price differentials, which are driven by different economic variables. By analyzing the factor loadings of the nonstationary latent factor, we identify some trading-friction and information-friction variables that have effects on the price convergence between the A- and H-shares.

The importance of the positional probability of word final (but not word initial) characters for word segmentation and identification in children and adults' natural Chinese reading.

Word spacing is important in guiding eye movements during spaced alphabetic reading. Chinese is unspaced and it remains unclear as to how Chinese readers segment and identify words in reading. We conducted two parallel experiments to investigate whether the positional probabilities of the initial and the final characters of a multicharacter word affected word segmentation and identification in Chinese reading. Two-character words were selected as targets. In Experiment 1, the initial character's positional probability was manipulated as being either high or low, and the final character was kept identical across the two conditions. In Experiment 2, an analogous manipulation was made for the final character of the target word. We recorded adults' and children's eye movements when they read sentences containing these words. In Experiment 1, reading times on targets did not differ in the two conditions for both children and adults, providing no evidence that a word initial character's positional probability contributes to word segmentation. In Experiment 2, adults had shorter reading times and made fewer refixations on targets that comprised final characters with high relative to low positional probabilities; a similar effect was observed in children, but this effect had a slower time course. The results demonstrate that the positional probability of the final (but not the initial) character of a word influences segmentation commitments in reading. It suggests that Chinese readers identify where a currently fixated word ends, and via this commitment, by default, they identify where the subsequent word begins. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

КСЕНОЛОГІЯ ЖІНОЧОГО: УКРАЇНСЬКІ ЖІНКИ В УЯВЛЕННЯХ КИТАЙСЬКОЇ МОЛОДІ

The article represents results of a socio-philosophical study of the ideas of educated Chinese youth about modern Ukrainian women. The survey had been carried out on Summer 2023 during a full-scale Russian war against Ukraine. At the same time, we proceeded from the standpoint of the current state of Ukrainian-Chinese relations and the least possible politicization of the survey. The study involved an analysis of associations and was an in-depth loosely coupled interview with a orientative set of questions. The guide-interview was written in Ukrainian and English, with subsequent translation into Chinese. It is important for research that the interviewer was a woman. The interview was conducted by the Chinese co-author of this article under usual conditions. The answers were translated into Ukrainian. Direct translation of some answers was quite difficult, so clarifications and explanations were made in English. The research is socio-philosophical and does not fully correspond to sociological canons. The sources of the formation of these ideas were also investigated. In particular, coverage of the status of Ukrainian women during the Russian war in the Chinese segment of the Internet in English and Chinese. The article demonstrates how the publications available on the sites outline ideas about modern women and then broadcasted by young people in the form of concepts, lines of thought, contexts and accents. By definition of xenology the author means the direction of socio-philosophical searches aimed at understanding the "outlying", "other" and "different". In the course of the research, it was found that the characteristics of Ukrainian women do not contain signs of "alien" or "enemy". It has been demonstrated that the socio-cultural space of Chinese youth is permeated with various stereotypes, among which gender stereotypes occupy a leading place. The need for further research into an important problem for Chinese youth – surrogate motherhood – had been well-founded. It had been noted that the study of stereotypes requires a methodology that uses intersectionality and is based on an interdisciplinary approach. The study had confirmed that socio-philosophical analysis based on an intersectional approach can serve the interdisciplinary transfer of experimental data in gender studies.

Research on the Innovation-Driving Mechanism for the Synergistic Development of Two-Way FDI in China’s Manufacturing Industry: Based on the Perspective of the New Development Pattern of “Dual Circulation”

The new pattern of “dual circulation” is a new development model for China to seek mutual promotion between international and domestic markets in the new era. In this context, this paper explores the synergistic relationship between two-way FDI and its impact mechanism on the improvement of China’s manufacturing innovation capability. By taking 27 segments of the Chinese manufacturing industry as data samples from 2003 to 2018, we use the Granger cause, orthogonalized impulse response function, and physical capacity coupling system to verify the two-way FDI synergistic development relationship in the Chinese manufacturing industry and measure its degree of synergy. In addition, we empirically explore the mediating role of industrial structure upgrading in the impact of two-way FDI synergistic development on innovation capability in the Chinese manufacturing industry, and further analyze the industrial heterogeneity of this mediating role among labor-intensive, capital-intensive, and technology-intensive manufacturing segments. The study finds that firstly, there are different degrees of synergistic development in the Chinese manufacturing segments, and this synergistic development significantly contributes to the innovation capability and industrial structure upgrading of the Chinese manufacturing industry. Secondly, industrial structure upgrading of the Chinese manufacturing industry plays an essential mediating effect in the innovation-driving process of the two-way FDI synergistic development, and the mediating effect shows significant industrial heterogeneity. More specifically, the mediating effects in labor-intensive and technology-intensive industries are significantly positive, and the mediating effect in technology-intensive industries is more prominent. However, the mediating effect in capital-intensive industries is significantly negative. The paper provides empirical evidence to clarify the innovation-driving mechanism of the two-way FDI synergistic development in China’s manufacturing industry from the perspective of the new development pattern of “dual circulation” and provides valuable references for research in related fields.