Chengxia Liu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Chengxia Liu has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
Lan Yao
School of Information Science and Engineering (School of Cyber Science and Technology), Zhejiang Sci-Tech University
2 shared publicationsFnag Yu
Yuanpei College, Shaoxing University, Shaoxing, Zhejiang, 312000, China
1 shared publicationJianing Wang
School of Fashion Design and Engineering, Zhejiang Sci-Tech University, Hangzhou, China
1 shared publicationJiawen Gu
School of Fashion Design and Engineering, Zhejiang Sci-Tech University, Hangzhou, China
1 shared publicationNingfeng Zhang
School of Fashion Design & Engineering , Zhejiang Sci-Tech University , Hangzhou , P.R. China
1 shared publication- Wenda Wei
School of Fashion Design and Engineering, Zhejiang Sci-Tech University, Hangzhou, China
1 shared publication
Latest Publications
- 6 May 2026
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 38 (3) , 581-597
Research on multi-label clothing image recognition based on deep learning
Purpose: Nowadays, the labeling of clothing images on e-commerce platforms mainly relies on manual labor, which not only wastes manpower and time, but also causes problems such as inconsistent labeling standards among merchants and cognitive differences among different annotators. Besides, it is particularly difficult to make image annotation for dresses with rich style variations and massive image data. To solve this problem, in this study, an automatic recognition method of multi-label images based on deep learning is proposed. Design/methodology/approach: In this paper, first dress images were selected whose attributes were annotated and a dataset for the dressing images was established according to the style of silhouette, collar, and sleeve. Then based on the baseline networks, a multi-label dress image recognition method, the VE-DE network, was proposed, which combines VGG-19, DenseNet-201 and the channel attention network. Findings: The results show that the average precision mean (mAP), precision, and recall of the dress images in the VE-DE recognition test set were 93.9%, 91.6%, and 90.3%, respectively, all higher than the three baseline networks. Among them, mAP was 4.3%, 11.1%, and 3.7% higher than VGG-19, ResNet-101, and DenseNet-201, respectively. VE-DE network can improve the vulnerability of the baseline network to the influence of external factors when identifying the profile and has a good recognition effect on the samples with easily confused profiles. Visual analysis of the class activation heatmap and intermediate layer activation of the VE-DE model shows that it can focus on such parts of the dress as the collar, sleeves, and waist, which contribute significantly to the recognition of the images, with the waist being the most highly focused area. Originality/value: The automatic recognition method of multi-label images can be used in an e-commerce platform. With the help of this technology, e-commerce platforms can automatically and quickly label clothing images, reduce labor costs, improve product labeling speed and quality.
- 6 May 2026
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 38 (3) , 581-597
Research on multi-label clothing image recognition based on deep learning
Purpose: Nowadays, the labeling of clothing images on e-commerce platforms mainly relies on manual labor, which not only wastes manpower and time, but also causes problems such as inconsistent labeling standards among merchants and cognitive differences among different annotators. Besides, it is particularly difficult to make image annotation for dresses with rich style variations and massive image data. To solve this problem, in this study, an automatic recognition method of multi-label images based on deep learning is proposed. Design/methodology/approach: In this paper, first dress images were selected whose attributes were annotated and a dataset for the dressing images was established according to the style of silhouette, collar, and sleeve. Then based on the baseline networks, a multi-label dress image recognition method, the VE-DE network, was proposed, which combines VGG-19, DenseNet-201 and the channel attention network. Findings: The results show that the average precision mean (mAP), precision, and recall of the dress images in the VE-DE recognition test set were 93.9%, 91.6%, and 90.3%, respectively, all higher than the three baseline networks. Among them, mAP was 4.3%, 11.1%, and 3.7% higher than VGG-19, ResNet-101, and DenseNet-201, respectively. VE-DE network can improve the vulnerability of the baseline network to the influence of external factors when identifying the profile and has a good recognition effect on the samples with easily confused profiles. Visual analysis of the class activation heatmap and intermediate layer activation of the VE-DE model shows that it can focus on such parts of the dress as the collar, sleeves, and waist, which contribute significantly to the recognition of the images, with the waist being the most highly focused area. Originality/value: The automatic recognition method of multi-label images can be used in an e-commerce platform. With the help of this technology, e-commerce platforms can automatically and quickly label clothing images, reduce labor costs, improve product labeling speed and quality.
- 5 December 2024
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 37 (1) , 138-153
Research on embroidery style migration model based on texture cycle GAN
Purpose: As an ancient art form, embroidery has strong practicality and artistic value. However, current embroidery style migration models produce images with unclear textures and a lack of stitch detail. So, in this paper, we propose a cyclic consistent embroidery style migration network with texture constraints, which is called Texture Cycle GAN (TCGAN).Design/methodology/approach: The model is based on the existing Cycle GAN network with an additional texture module. This texture module is implemented using a pre-trained Markovian adversarial network to synthesize embroidery texture features. The overall algorithm consists of two generative adversarial networks (for style migration) and the Markovian adversarial network (for texture synthesis).Findings: Qualitative and quantitative experiments show that, compared with the existing convolutional neural network style transfer algorithm, the introduction of the texture-constrained embroidery style transfer model TCGAN can effectively learn the characteristics of style images, generate digital embroidery works with clear texture and natural stitches and achieve more realistic embroidery simulation effects.Originality/value: By improving the algorithm for image style migration and designing a reasonable loss function, the generated embroidery patterns are made more detailed, which shows that the model can improve the realism of embroidery style simulation and help to improve the standard of embroidery craftsmanship, thus promoting the development of the embroidery industry.
- 5 December 2024
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 37 (1) , 138-153
Research on embroidery style migration model based on texture cycle GAN
Purpose: As an ancient art form, embroidery has strong practicality and artistic value. However, current embroidery style migration models produce images with unclear textures and a lack of stitch detail. So, in this paper, we propose a cyclic consistent embroidery style migration network with texture constraints, which is called Texture Cycle GAN (TCGAN).Design/methodology/approach: The model is based on the existing Cycle GAN network with an additional texture module. This texture module is implemented using a pre-trained Markovian adversarial network to synthesize embroidery texture features. The overall algorithm consists of two generative adversarial networks (for style migration) and the Markovian adversarial network (for texture synthesis).Findings: Qualitative and quantitative experiments show that, compared with the existing convolutional neural network style transfer algorithm, the introduction of the texture-constrained embroidery style transfer model TCGAN can effectively learn the characteristics of style images, generate digital embroidery works with clear texture and natural stitches and achieve more realistic embroidery simulation effects.Originality/value: By improving the algorithm for image style migration and designing a reasonable loss function, the generated embroidery patterns are made more detailed, which shows that the model can improve the realism of embroidery style simulation and help to improve the standard of embroidery craftsmanship, thus promoting the development of the embroidery industry.
- 7 May 2024
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 36 (3) , 393-406
Research on the new method for visualizing bendability testing of fabrics in multiple directions
Purpose: The current bending test method can only test the bending performance of fabric in one direction at a time. It is not possible to directly observe the bending morphology of fabrics in different directions, and it is necessary to cut samples and repeat the test several times, which takes more time. For this situation, a multidirectional visualization of the fabric bending test method is proposed, using which multiple results can be obtained at one time and the fabric bending can be visualized.Design/methodology/approach: About 17 fabrics are tested using a self-designed device. The fabrics are cut into special triangles and multiple sets of results in three directions are obtained at once using the device. The experimental specimens are photographed from the above and the transverse elongation length, bending projection area and circumference are extracted after image processing.Findings: The results show that the correlation coefficients of transverse elongation, bending projected area and circumference are good with the bending length measured by the cantilever method. In which, all three indicators are positively correlated with the bending length. This indicates the good feasibility of the new method.Originality/value: This method can get the bending index of fabrics in three directions, with five samples in each direction at one time. Meanwhile, it can also visualize the flexural differences between different fabrics and directions of the same fabric. It can provide more efficient testing means for the textile testing field, and the testing efficiency is 15 times of the existing method, which has better theoretical significance and practical values.
- 7 May 2024
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 36 (3) , 393-406
Research on the new method for visualizing bendability testing of fabrics in multiple directions
Purpose: The current bending test method can only test the bending performance of fabric in one direction at a time. It is not possible to directly observe the bending morphology of fabrics in different directions, and it is necessary to cut samples and repeat the test several times, which takes more time. For this situation, a multidirectional visualization of the fabric bending test method is proposed, using which multiple results can be obtained at one time and the fabric bending can be visualized.Design/methodology/approach: About 17 fabrics are tested using a self-designed device. The fabrics are cut into special triangles and multiple sets of results in three directions are obtained at once using the device. The experimental specimens are photographed from the above and the transverse elongation length, bending projection area and circumference are extracted after image processing.Findings: The results show that the correlation coefficients of transverse elongation, bending projected area and circumference are good with the bending length measured by the cantilever method. In which, all three indicators are positively correlated with the bending length. This indicates the good feasibility of the new method.Originality/value: This method can get the bending index of fabrics in three directions, with five samples in each direction at one time. Meanwhile, it can also visualize the flexural differences between different fabrics and directions of the same fabric. It can provide more efficient testing means for the textile testing field, and the testing efficiency is 15 times of the existing method, which has better theoretical significance and practical values.
- 12 September 2023
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 35 (6) , 904-917
Novel measurement of fabric bending anisotropy
Purpose: Under the widely used testing methods for fabric bending behavior, only one result for one direction can be obtained by using one fabric which is low efficient. To obtain fabric bending anisotropy, it is necessary to conduct a great many testing experiments. The purpose of this paper is to investigate a novel, efficient and visual method that can measure fabric bending anisotropy.Design/methodology/approach: Fabrics are first cut into special shapes with eight strips including four directions, 0°(warp direction), 90°(weft direction), 45 and 135°(true bias), then are put onto the self-designed instrument. After that a camera is used to take picture from the right above the prism. New parameters, projection area, projection length, projection length, falling height and falling index (S, L, H and I in short) are obtained. Furthermore, standard deviation of them (SDS, SDL, SDH and SDI in short) are extracted for the characterization of bending anisotropy.Findings: Results show that the new method has good feasibility and S, L and I can be used to express fabric bending property well. Of all the four new parameters, SDL has the highest correlation with SD of bending length, SDS the second and SDH the third. That is, SDL can characterize bending anisotropy best. Taken convenience of data acquisition and correlation into consideration, bending length L is the best parameter. Average L and SDL in four directions can be combined to express the comprehensive bending behavior of fabrics.Originality/value: The new method can measure and characterize both the fabric bending property and bending anisotropy. Besides its high efficiency, it can display fabric bending or bending anisotropy visually and directly.
- 12 September 2023
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 35 (6) , 904-917
Novel measurement of fabric bending anisotropy
Purpose: Under the widely used testing methods for fabric bending behavior, only one result for one direction can be obtained by using one fabric which is low efficient. To obtain fabric bending anisotropy, it is necessary to conduct a great many testing experiments. The purpose of this paper is to investigate a novel, efficient and visual method that can measure fabric bending anisotropy.Design/methodology/approach: Fabrics are first cut into special shapes with eight strips including four directions, 0°(warp direction), 90°(weft direction), 45 and 135°(true bias), then are put onto the self-designed instrument. After that a camera is used to take picture from the right above the prism. New parameters, projection area, projection length, projection length, falling height and falling index (S, L, H and I in short) are obtained. Furthermore, standard deviation of them (SDS, SDL, SDH and SDI in short) are extracted for the characterization of bending anisotropy.Findings: Results show that the new method has good feasibility and S, L and I can be used to express fabric bending property well. Of all the four new parameters, SDL has the highest correlation with SD of bending length, SDS the second and SDH the third. That is, SDL can characterize bending anisotropy best. Taken convenience of data acquisition and correlation into consideration, bending length L is the best parameter. Average L and SDL in four directions can be combined to express the comprehensive bending behavior of fabrics.Originality/value: The new method can measure and characterize both the fabric bending property and bending anisotropy. Besides its high efficiency, it can display fabric bending or bending anisotropy visually and directly.
- 18 April 2023
- Published by Taylor & Francis in The Journal of the Textile Institute
- Vol. 115 (2) , 259-265
New method for measuring fabric multi-directional stiffness and drape
A method that can measure fabric multi-directional bending performance and draping property simultaneously is proposed and new parameters, projection area and projection length are extracted for the characterization. A total of 30 fabrics are tested with the new method as well as with the cantilever test and umbrella method. Results show that the new parameters have good correlationship with bending performance and draping property. Bending performance in four directions and two groups of bending indexes in each direction can be acquired with only one fabric sample. Compared with the conventional cantilever method widely used, the method proposed in this article is more efficient and time-saving. At the same time, it can test fabric drape in one specific direction. Therefore, it can combine the test of fabric bending and draping performance into one.
- 18 April 2023
- Published by Taylor & Francis in The Journal of the Textile Institute
- Vol. 115 (2) , 259-265
New method for measuring fabric multi-directional stiffness and drape
A method that can measure fabric multi-directional bending performance and draping property simultaneously is proposed and new parameters, projection area and projection length are extracted for the characterization. A total of 30 fabrics are tested with the new method as well as with the cantilever test and umbrella method. Results show that the new parameters have good correlationship with bending performance and draping property. Bending performance in four directions and two groups of bending indexes in each direction can be acquired with only one fabric sample. Compared with the conventional cantilever method widely used, the method proposed in this article is more efficient and time-saving. At the same time, it can test fabric drape in one specific direction. Therefore, it can combine the test of fabric bending and draping performance into one.
- 20 December 2022
- Published by Springer Nature in Fibers and Polymers
- Vol. 23 (13) , 3638-3644
Novel Method for Measurement of Fabric Multi-directional Bending Performance
- 20 December 2022
- Published by Springer Nature in Fibers and Polymers
- Vol. 23 (13) , 3638-3644
Novel Method for Measurement of Fabric Multi-directional Bending Performance
- Open Access
- 1 December 2022
- Published by Walter de Gruyter GmbH in Fibres & Textiles in Eastern Europe
- Vol. 30 (6) , 80-90
Study on the Influence of Elastic Compression Pants Elasticity and Movement Speed on Human Joint Protection
The aim of this study was to clarify how human movement speed and pants elasticity affect the athletic performance of knee joint protection by testing pants with different elastic properties: CP1 (tight pants) and CP2&CP3 (elastic compression pants), which reinforce the knee joint. In addition, CS (cotton sport pants) was developed as a control garment. Three subjects wore CP1, CP2, CP3 and CS while running on the treadmill at three kinds of human movement speed. A three-dimensional motion capture instrument was used to capture the three-dimensional trajectory of the marked points of lower limbs. As a result, the influence of the movement speed on the kinematic parameters (AKJ & gait cycle) was more obvious than the fabric elasticity. If elastic pants are worn during running, the change of AKJ will be stable with the increase of speed. When non-elastic pants are worn, the effect is opposite. Not only that, elastic compression pants are efficient in reducing the motion amplitude of the knee joint during the suspension period as far as 41°, making it highly practical in terms of stability. That is, the elastic fabric can protect the joints when the lower limbs are in motion. Moreover, with the increase of speed and elasticity, the elastic pants can reduce the gait cycle by up to 22% compared with non-elastic pants alone. Through the kinematics mechanism of human joints, these findings may translate into an effect on protective performance and a reduction in sport injuries. Therefore, it is necessary to wear elastic pants, especially compression pants, when running at higher speed, as the average gait cycle gradually decreases. This research shows that the knee joint protection functions of elastic compression garments differ according to the level of elasticity and differential movement speed, providing theoretical support for designing and producing elastic compression pants. It also acts as a guide for the research of lower limb joint protection.
- 1 December 2022
- Published by Walter de Gruyter GmbH in Fibres & Textiles in Eastern Europe
- Vol. 30 (6) , 80-90
Open AccessStudy on the Influence of Elastic Compression Pants Elasticity and Movement Speed on Human Joint Protection
The aim of this study was to clarify how human movement speed and pants elasticity affect the athletic performance of knee joint protection by testing pants with different elastic properties: CP1 (tight pants) and CP2&CP3 (elastic compression pants), which reinforce the knee joint. In addition, CS (cotton sport pants) was developed as a control garment. Three subjects wore CP1, CP2, CP3 and CS while running on the treadmill at three kinds of human movement speed. A three-dimensional motion capture instrument was used to capture the three-dimensional trajectory of the marked points of lower limbs. As a result, the influence of the movement speed on the kinematic parameters (AKJ & gait cycle) was more obvious than the fabric elasticity. If elastic pants are worn during running, the change of AKJ will be stable with the increase of speed. When non-elastic pants are worn, the effect is opposite. Not only that, elastic compression pants are efficient in reducing the motion amplitude of the knee joint during the suspension period as far as 41°, making it highly practical in terms of stability. That is, the elastic fabric can protect the joints when the lower limbs are in motion. Moreover, with the increase of speed and elasticity, the elastic pants can reduce the gait cycle by up to 22% compared with non-elastic pants alone. Through the kinematics mechanism of human joints, these findings may translate into an effect on protective performance and a reduction in sport injuries. Therefore, it is necessary to wear elastic pants, especially compression pants, when running at higher speed, as the average gait cycle gradually decreases. This research shows that the knee joint protection functions of elastic compression garments differ according to the level of elasticity and differential movement speed, providing theoretical support for designing and producing elastic compression pants. It also acts as a guide for the research of lower limb joint protection.
- 30 April 2022
- Published by The National Research and Development Institute for Textiles and Leather in Industria Textila
- Vol. 73 (02) , 204-212
Development of a hat style recognition system based on image processing and machine learning
- 30 April 2022
- Published by The National Research and Development Institute for Textiles and Leather in Industria Textila
- Vol. 73 (02) , 204-212
Development of a hat style recognition system based on image processing and machine learning
- 28 October 2021
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 33 (5) , 859-875
Objective evaluation of optical illusion skirt based on image texture features
Purpose: Nowadays, most methods of illusion garment evaluation are based on the subjective evaluation of experienced practitioners, which consumes time and the results are too subjective to be accurate enough. It is necessary to explore a method that can quantify professional experience into objective indicators to evaluate the sensory comfort of the optical illusion skirt quickly and accurately. The purpose of this paper is to propose a method to objectively evaluate the sensory comfort of optical illusion skirt patterns by combining texture feature extraction and prediction model construction.Design/methodology/approach: Firstly, 10 optical illusion sample skirts are produced, and 10 experimental images are collected for each sample skirt. Then a Likert five-level evaluation scale is designed to obtain the sensory comfort level of each skirt through the questionnaire survey. Synchronously, the coarseness, contrast, directionality, line-likeness, regularity and roughness of the sample image are calculated based on Tamura texture feature algorithm, and the mean, contrast and entropy are extracted of the image transformed by Gabor wavelet. Both are set as objective parameters. Two final indicators T1 and T2 are refined from the objective parameters previously obtained to construct the predictive model of the subjective comfort of the visual illusion skirt. The linear regression model and the MLP neural network model are constructed.Findings: Results show that the accuracy of the linear regression model is 92%, and prediction accuracy of the MLP neural network model is 97.9%. It is feasible to use Tamura texture features, Gabor wavelet transform and MLP neural network methods to objectively predict the sensory comfort of visual illusion skirt images.Originality/value: Compared with the existing uncertain and non-reproducible subjective evaluation of optical illusion clothing based on experienced experts. The main advantage of the authors' method is that this method can objectively obtain evaluation parameters, quickly and accurately obtain evaluation grades without repeated evaluation by experienced experts. It is a method of objectively quantifying the experience of experts.
- 28 October 2021
- Published by Emerald Publishing in International Journal of Clothing Science and Technology
- Vol. 33 (5) , 859-875
Objective evaluation of optical illusion skirt based on image texture features
Purpose: Nowadays, most methods of illusion garment evaluation are based on the subjective evaluation of experienced practitioners, which consumes time and the results are too subjective to be accurate enough. It is necessary to explore a method that can quantify professional experience into objective indicators to evaluate the sensory comfort of the optical illusion skirt quickly and accurately. The purpose of this paper is to propose a method to objectively evaluate the sensory comfort of optical illusion skirt patterns by combining texture feature extraction and prediction model construction.Design/methodology/approach: Firstly, 10 optical illusion sample skirts are produced, and 10 experimental images are collected for each sample skirt. Then a Likert five-level evaluation scale is designed to obtain the sensory comfort level of each skirt through the questionnaire survey. Synchronously, the coarseness, contrast, directionality, line-likeness, regularity and roughness of the sample image are calculated based on Tamura texture feature algorithm, and the mean, contrast and entropy are extracted of the image transformed by Gabor wavelet. Both are set as objective parameters. Two final indicators T1 and T2 are refined from the objective parameters previously obtained to construct the predictive model of the subjective comfort of the visual illusion skirt. The linear regression model and the MLP neural network model are constructed.Findings: Results show that the accuracy of the linear regression model is 92%, and prediction accuracy of the MLP neural network model is 97.9%. It is feasible to use Tamura texture features, Gabor wavelet transform and MLP neural network methods to objectively predict the sensory comfort of visual illusion skirt images.Originality/value: Compared with the existing uncertain and non-reproducible subjective evaluation of optical illusion clothing based on experienced experts. The main advantage of the authors' method is that this method can objectively obtain evaluation parameters, quickly and accurately obtain evaluation grades without repeated evaluation by experienced experts. It is a method of objectively quantifying the experience of experts.