Review

Artificial intelligence and sustainability in the fashion industry:

a review from 2010 to 2022
Leo Ramos1,4 · Francklin Rivas‑Echeverría1,2 · Anna Gabriela Pérez3 · Edmundo Casas4

Received: 15 August 2023 / Accepted: 8 November 2023

© The Author(s) 2023  OPEN

Abstract
The fashion industry often falls short of sustainability goals, but contemporary technological advancements offer a wide
range of tools to address this issue. Artificial Intelligence (AI) has emerged as a particularly promising ally in promoting
sustainability in fashion. This literature review explores how AI can contribute to the fashion industry’s sustainability,
highlighting its potential benefits and limitations. Following PRISMA guidelines, we conducted a review of scientific
documents, focusing on the period from 2010 to 2022. After a meticulous selection process, we analyzed 37 scholarly
articles to distill their key insights and contributions. Our findings demonstrate that AI has diverse applications in different
aspects of the fashion industry, enhancing sustainability efforts in supply chain management, creative design, sales and
promotion, waste control, and data analysis. While AI offers significant potential, it is important to acknowledge limita‑
tions, such as the volume of data required and associated implementation costs. The reviewed literature aligns with the
multifaceted nature of sustainability, emphasizing responsible resource management, accessible services, and efficient
customer satisfaction, both now and in the future. In conclusion, despite some reservations, AI stands as a crucial partner
in guiding the fashion industry toward a more sustainable future.

Article Highlights

• AI presents promising that can be applied to design, • Many of the research studied focus on environmental
sales, and wastemanagement, all aimed at boosting dimension, includingecofriendly manufacturing and
sustainability. defect detection.
• Despite reluctance, AI is becoming a vital ally in moving
the fashion industrytoward sustainability.

Keywords Artificial intelligence · Fashion industry · Sustainability · Sustainable fashion · Sustainable development

* Leo Ramos, [email protected]; Francklin Rivas‑Echeverría, [email protected]; Anna Gabriela Pérez, gabipm23@
gmail.com; Edmundo Casas, [email protected] | 1School of Mathematical and Computational Sciences, Yachay Tech University,
Urcuquí 100119, Imbabura, Ecuador . 2Pontificia Universidad Católica del Ecuador Sede Ibarra, Ibarra 10102, Imbabura, Ecuador
. 3Laboratorio de Sistemas Inteligentes, Universidad de Los Andes, Mérida 05101, Venezuela . 4Kauel Inc., Houston, TX 77027, USA.

SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

Vol.:(0123456789)
Review SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

1 Introduction tools that will enable it to achieve sustainability goals.

As a result, the fashion industry is increasingly turning
Sustainability, as defined in Ref. [1], entails meeting pre‑ to AI to help improve sustainability [14]. AI is a branch
sent needs without endangering the requirements of of computing that develops systems for simulating the
future generations. It involves safeguarding the natu‑ cognitive capabilities of humans, especially in problem-
ral environment and social well-being while ensuring solving tasks [15]. AI is a trending area, and its use has
that economic growth and development do not pose spread to multiple areas, such as medicine, science, and
threats. Sustainability is a critical concern in the industry industry.
[2]. Today, it is imperative to integrate social and envi‑ Through AI, the fashion industry can optimize vari‑
ronmental considerations into business decision-making ous processes in the apparel production process [16, 17].
and operations [3]. In specific industries, including agri‑ Moreover, AI algorithms enable companies to use their
culture, mining, renewable energy, and manufacturing, resources better, leading to cost reduction, increased effi‑
sustainability has emerged as a central objective aimed ciency and effectiveness, and increased production speed
at mitigating adverse impacts on the environment and [17]. In addition, this benefits the environment and soci‑
society [4]. ety as AI makes it possible to process and use natural and
Sustainability comprises three key dimensions: eco‑ human resources better [18].
nomic, social, and environmental, as noted by Kristensen This study provides a focused examination of how AI
[5]. The economic dimension ensures that production can enhance sustainability within the fashion industry. We
meets present needs without compromising future capac‑ concentrate on evaluating specific areas where AI can be
ity [5, 6]. The social dimension focuses on parameters for applied in fashion, as well as assessing the performance
social equity, access to essential services, security, and of different AI techniques in bolstering sustainability. Our
citizen participation in governance [7]. The environmen‑ primary aim is to discern both the merits and drawbacks
tal dimension emphasizes responsible resource manage‑ associated with these AI approaches, offering valuable
ment and waste control to prevent overconsumption and insights for experts and stakeholders in these fields.
environmental degradation [5, 8]. These dimensions are
interconnected and integral to sustainable development. 2 Methodology
The United Nations argues that strategies to gener‑
ate economic growth must go hand in hand with strate‑ This systematic review was conducted using the Preferred
gies to promote prosperity and protect the planet [9]. Reporting Items for Systematic Reviews and Meta-Analy‑
These include a range of social needs such as education, ses (PRISMA) statements [19], the most commonly used
health, and job opportunities, while at the same time reporting guidelines for systematic reviews [20]. The fol‑
ensuring that climate change is halted and the environ‑ lowing is a description of the stages used to carry out this
ment is protected. work.
The fashion industry, notorious for its sustainability
shortcomings [10], consumes vast resources in clothing 2.1 Research questions
and accessory production, particularly straining water
resources. It ranks among the largest water consumers, This article aims to explore how AI is used in the fashion
often with inadequate treatment post-use [2]. Countries industry to improve sustainability and answer the follow‑
like Bangladesh face issues of heavy metal and micro‑ ing questions:
plastic pollution in water sources [11, 12], leading to
health problems among nearby residents who consume Q1 In which ways can artificial intelligence improve sus‑
this water [13]. tainability in the fashion industry?
Furthermore, the fashion industry is marred by sig‑ Q2 What are the advantages of using artificial intelli‑
nificant labor exploitation, often subjecting workers to gence as a tool to achieve sustainability?
grueling shifts exceeding eight hours a day. Additionally, Q3 What are the limitations of using artificial intelligence
the transportation of clothing and accessories further as a tool to achieve sustainability?
exacerbates greenhouse gas emissions. Moreover, the
fast fashion model has accelerated the production and 2.2 Eligibility criteria
disposal of clothing, resulting in a surge in waste gen‑
eration and unsustainable resource consumption [10]. 2.2.1 Inclusion criteria
As a countermeasure to all these problems, the fash‑
ion industry is increasingly trying to find solutions and Regarding the inclusion criteria (IC), an article was
included as long as it met all of the following criteria:

Vol:.(1234567890)
SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2 Review

Table 1  Online sources used in this work 2.5 Selection process

No. Source Url
From the preselected articles, titles and abstracts were
1 Scopus www.​scopus.​com imported into Covidence1 systematic review software
2 ScienceDirect www.​scien​cedir​ect.​com for screening. First, using Covidence tools, duplicate
3 ACM Digital Library https://​dl.​acm.​org articles were identified and removed. Then, the articles
4 IEEE Xplore https://​ieeex​plore.​ieee.​org were manually reviewed by two researches (LR and FR) to
5 Springer Link www.​link.​sprin​ger.​com remove any remaining duplicates. Next, all the research‑
6 Google Scholar https://​schol​ar.​google.​com/ ers independently screened the titles and abstracts of
the articles. In case of disagreement, the consensus was
reached to determine articles to screen in the next stage
IC1 Empirical research, not books, manuals, or tutorials. by discussion.
IC2 Research that explicitly uses AI techniques as a The next step was to remove articles that could not be
potential tool to drive sustainability in the fashion accessed. That is, articles whose full-text was unavailable
industry. were removed and not included for full-text screening.
IC3 Research published between 2010 and 2022. This was done by one researcher (LR).
IC4 Research published in peer-reviewed journals only. Then, the next step consisted of retrieving the articles
and screening them by full-text reading. In this step, the
2.2.2 Exclusion criteria inclusion and exclusion criteria were considered to deter‑
mine which articles passed to the next stage and which
Regarding exclusion criteria (EC), an article was excluded did not. This was done by two researchers (LR and FR) and
if it failed to meet any of the following criteria: verified by the other (AGP and EC).
Finally, the articles that met the inclusion criteria and
EC1 Research that does not involve approaches based evidenced a relevant contribution to the objectives of this
on AI. study were included in the review.
EC2 Research not published in a peer-reviewed journal.
EC3 Research not written in English. 2.6 Data collection process
EC4 Document not available.
A data extraction sheet was developed for this stage. The
2.3 Information sources first version of the extraction sheet was made by two
researchers (LR and FR). This was first tested with ten ran‑
We decided to use multiple databases and search engines domly selected articles. Subsequently, the other research‑
to expand the number of relevant articles considered. ers (AGP and EC) verified and validated that the extraction
Details of these are given in Table 1. sheet worked correctly and allowed all relevant informa‑
Other sources, such as trial registers or other grey litera‑ tion to be obtained. They also made corrections to the
ture sources, were not used. extraction sheet when necessary.

2.4 Search strategy 2.7 Information extraction

We performed a search string based on the previously To address our research questions, the articles selected for
mentioned inclusion criteria to search for articles in the review were thoroughly examined to extract the following
databases. This was: ( ( ( "sustainability" AND "fashion indus- main information:
try" ) OR "sustainable fashion" ) AND ( "artificial intelligence"
OR "machine learning" OR "deep learning" OR "expert sys- • Potential application domain or area.
tems" OR "knowledge-based systems" ) ). Likewise, we apply • AI class used.
a series of filters, where possible, in each of the databases • Aim of the work.
to obtain only relevant articles for our review. Details of • Main technique(s) used.
this can be seen in Table 2. The databases were searched • Relevant findings.
on December 30, 2022. • Publication year.

1
https://​www.​covid​ence.​org/.

Vol.:(0123456789)
Review SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

Table 2  Filters and specifications for searching online sources

Source Search string Filters

Scopus TITLE-ABS-KEY(("sustainability" AND "fashion industry") OR Article type: Research article

"sustainable fashion") AND TITLE-ABS-KEY("artificial intel‑ Year(s): 2010–2022
ligence" OR "machine learning" OR "deep learning" OR
"expert systems" OR "knowledge-based systems")
ScienceDirect (("sustainability" AND "fashion industry") OR "sustain‑ Article type: Research article
able fashion") AND ("artificial intelligence" OR "machine Year(s): 2010–2022
learning" OR "deep learning" OR "expert systems" OR
"knowledge-based systems")
ACM Digital Library [[[All: "sustainability"] AND [All: "fashion industry"]] OR [All: Content type: Research article
"sustainable fashion"]] AND [[All: "artificial intelligence"] Publication: Journals
OR [All: "machine learning"] OR [All: "deep learning"] Publication date: 2010–2022
OR [All: "expert systems"] OR [All: "knowledge-based
systems"]]
IEEE Xplore ((("All Metadata":sustainability AND "All Metadata":fashion Type: Journals
industry) OR "All Metadata":sustainable fashion) Year(s): 2010–2022
AND ("All Metadata":artificial intelligence OR "All
Metadata":machine learning OR "All Metadata":deep
learning OR "All Metadata":expert systems OR "All
Metadata":knowledge-based systems))
Springer Link (("sustainability" AND "fashion industry") OR "sustain‑ Content type: Article
able fashion") AND ("artificial intelligence" OR "machine Date published: 2010–2022
learning" OR "deep learning" OR "expert systems" OR Language: English
"knowledge-based systems")
Google Scholar (("sustainability" AND "fashion industry") OR "sustain‑ Exclude: Patents
able fashion") AND ("artificial intelligence" OR "machine Date published: 2010–2022
learning" OR "deep learning" OR "expert systems" OR Without the words: “review", “meta-analysis", “systematic
"knowledge-based systems") review", “exploratory study", “bibliometric review",
“literature review", “systematic literature review"

In a complementary manner, the publishing company and Given that we studied different domains in this work, a
the journal in which each article was published were also wide variety of journals host this type of research. Table 4
extracted. The information extraction was carried out and below highlights the top four journals among those that
cross-checked by all the researchers in this work. contributed significantly to this study.
From this table, the Journal of Cleaner Production stands
out as the journal that contributes the most to this study,
3 Results providing 10.81% of the articles reviewed. To integrate
the textile industry with technology, the International
We identified 616 articles from the selected online sources. Journal of Clothing Science and Technology is the journal
Articles were screened and selected, as shown in Fig. 1, that makes the second most significant contribution, with
resulting in 37 studies meeting our inclusion criteria and 8.11%. This could be due to the significant development
ultimately being included in the review. of technology and its irruption in different industries in
Figure 2 shows the distribution of relevant articles recent years. Then, the journals Multimedia Tools and Appli-
retrieved per year. From this graph, it can be seen that from cations and Textile Research Journal contribute 5.41% of the
2019 onwards, there has been an increase in sustainability, articles. Finally, all the other journals have a contribution
the fashion industry, and AI research. Most of the research of 2.7% each.
is focused on the year 2020. The year 2022 presents a low Subsequently, we categorized the articles based on
number of articles since most of the articles we could not their potential applications in promoting sustainability
access are from this year. Concerning the publishers, it is within the fashion industry. The allocation of each arti‑
evident that Elsevier is the academic publishing company cle to a specific domain was achieved through extensive
that contributes most to this work. It provides more than discussions and consensus among all the researchers
40% of the articles retrieved, as shown in Table 3. In sec‑ involved in this study. The findings reveal that a majority of
ond place is Springer, followed by the other publishing the reviewed articles focus on optimizing the supply chain,
companies. comprising 40.54% of the total articles. Following closely,

Vol:.(1234567890)
SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2 Review

Fig. 1  PRISMA flow diagram

used in this work

27.03% of the articles are dedicated to sustainable cloth‑ most frequently employed. Figure 3 illustrates the prevalent
ing design and sales. Reducing waste represents 18.91% of use of ANNs in various forms, including traditional multi-
the articles, while data analysis contributes 13.52%. These layer feed-forward neural networks, convolutional neural
insights are summarized in Table 5. networks (CNNs) for image analysis, and generative adver‑
Upon closer examination of the technologies used in the sarial networks (GANs) for image generation. Additionally,
retrieved articles, artificial neural networks (ANNs) are the fuzzy logic and classical machine learning (ML) algorithms

Vol.:(0123456789)
Review SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

like k-means, random forests, and support vector machines

are commonly featured. Other technologies, such as block‑
chain and the Internet of Things (IoT), are also utilized.
We also classified the articles based on the sector to
which their contributions are oriented, including govern‑
ment, business, or customer orientation. Similarly, we cat‑
egorized each article according to the sustainability dimen‑
sion that best represents its contribution. In cases where an
article could belong to more than one dimension or target
sector, we selected the most representative one through
consensus among all the researchers involved in this study.
Fig. 2  Overall distribution of retrieved articles over time The results regarding the sustainability dimension are
summarized in Table 6. These findings reveal that the major‑
ity of the articles are situated within the economic dimen‑
Table 3  Distribution of articles based on publisher sion, focusing primarily on measures aimed at satisfying the
Publisher Amount Percent (%)
present and future needs of customers. Specifically, 41% of
the reviewed articles fall within this dimension. Following
Elsevier 16 43.24 closely is the environmental dimension, encompassing 35%
Springer 7 18.91 of the articles. This sector is notable for its contributions
Emerald 3 8.11 related to waste management, control, and reduction, with
MDPI 3 8.11 several works dedicated to enhancing recycling practices.
SAGE 2 5.41 Lastly, the social dimension comprises 24% of the articles,
Wiley 2 5.41 driven by contributions aimed at improving accessibility and
Taylor and Francis 2 5.41 services
IEEE 1 2.7 Regarding the target sector, the outcomes of this clas‑
Growing Science 1 2.7 sification are detailed in Table 7. The data indicates that the
Total 37 100 majority of the articles are business-oriented (84%), empha‑
sizing sustainability measures implemented within company
operations. These encompass actions concerning the sup‑
Table 4  Four leading journals contributing to this review ply chain, manufacturing processes, and design, all aimed at
Journal Amount Percent (%)
enhancing sustainability. In the second position is the cus‑
tomer sector (16%), focusing on initiatives aimed at improv‑
Journal of Cleaner Production 4 10.81 ing the customer’s shopping experience while encouraging
International Journal of Clothing Sci‑ 3 8.11 the consumption of sustainable products. Conversely, the
ence and Technology
government sector exhibits minimal relevance in this con‑
Multimedia Tools and Applications 2 5.41
text (0%).
Textile Research Journal 2 5.41
The summary of selected articles in terms of their main
characteristics is presented in Table 8. This data extraction
table serves as a comprehensive guide to understanding the
Table 5  Distribution of articles based on domain of application landscape of the research included in this review. It details
Domain Amount Percent (%)
key aspects such as the year of publication, methodologies
employed, objectives, and key findings. This consolidation of
Supply chain optimization 15 40.54 information is designed to offer readers an easily navigable
Design and sale of sustainable 10 27.03 overview, thereby facilitating a deeper comprehension of
clothing the review’s scope, methods, and results.
Reducing waste 7 18.91
Data analysis 5 13.52
Total 37 100 4 Discussion

From the reviewed articles, it was possible to highlight four

main application areas in which AI can help to improve
sustainability in the fashion industry. Each of these is
described below.

Vol:.(1234567890)
SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2 Review

Fig. 3  Distribution of articles

by AI used over time. It was
selected the most relevant
technology used in each
retrieved article

Table 6  Articles according to the sustainability dimension assigned processing to manufacturing, distribution, and final
to each product delivery, all while promoting sustainability [62].
Sustainability dimension Amount Percent‑ References Numerous works have already incorporated AI to opti‑
age (%) mize supply chains. For instance, in Refs. [31, 30], ML and
ANNs, respectively, are proposed for classifying cloth‑
Economic 15 41 [6, 21–34]
ing categories. Both approaches achieved high accuracy
Environmental 13 35 [35–47]
rates (> 80%), correctly categorizing clothing and sub‑
Social 9 24 [48–56]
categories. Models like these enable fashion companies
to automate classification tasks, ensuring organized and
efficient product category management.
Table 7  Articles according to the target sector assigned to each In the manufacturing domain [36], suggests using ML,
Target sector Amount Percentage References
specifically, support vector machine to detect common
(%) fabric defects, such as neps, broken ends, broken picks,
and oil stains from images, obtaining high accuracy (>
Business 31 84 [6, 21–48, 50, 57] 98%). Similarly, in Ref. [33], CNNs are employed for color
Customer 6 16 [49, 51, 52, 54–56] difference detection, a common defect in warp-knitted
Government 0 0 – fabrics. The authors utilized the YOLO neural architecture
and achieved real-time accuracy. These examples illus‑
trate AI’s positive impact on sustainable apparel manu‑
4.1 Supply chain optimization facturing, enhancing efficiency and resource optimiza‑
tion by identifying manufacturing faults and preventing
The supply chain encompasses the entire process, from defective garments from reaching consumers, ultimately
the creation of a product or service to its delivery to the reducing waste.
end consumer. It can be visualized as a network [59] com‑ Continuing with manufacturing [52], proposes a ML-
prising human and material components, all striving to based system that combines dimensionality reduction
minimize costs and maximize efficiency without compro‑ techniques and k-means based on 3D scans to define
mising the final product or service’s quality. However, real- adaptive morphotype mannequins. This innovative
world supply chains encounter numerous challenges [57], approach eliminates the need for predefined tables,
including shortages, delayed deliveries, and difficulties in ensuring garments better fit customers’ shapes, thus
adapting to changing market demands [57, 60]. optimizing manufacturing resources.
Sustainability has become a central theme in supply AI can also be utilized to select the best components
chains, especially within the fashion industry. However, for garments. For example, in Ref. [27], an expert sys‑
many supply chains in this sector still need to become tem is developed to identify and select the best type
more sustainable [61]. AI offers a myriad of benefits to of cotton fiber for product creation, leveraging docu‑
supply chains that would be otherwise unattainable mented knowledge sources and customer experience.
[57]. It provides a range of tools applicable throughout The system optimizes clothing manufacturing, reducing
the supply chain, from procurement and raw material

Vol.:(0123456789)
 Table 8  Summary of reviewed articles
References Domain AI class Aim Main technique Key findings Year
Review

[21] Data analysis Machine learning To perform customer segmenta‑ K-medoids Six market segments and 49 2015
tion to obtain market segments rules that allowed a better

Vol:.(1234567890)
and characterize subgroups of understanding of customer
observations. preferences in a highly custom‑
ized fashion manufacturer/e-
tailor were obtained.
[35] Data analysis Machine learning Evaluate the Brazilian consumer’s J48 It was found that the majority of 2019
ecological footprint and the Brazilian consumers surveyed
awareness of sustainability in wear cotton clothing but are
the clothing production stages. not aware of how it is manu‑
factured. Also, most consumers
seek to buy sustainable prod‑
ucts, but do not know if the
manufacturers are sustainable.
[48] Design and sale of sustainable Deep learning To propose an ANN model that Generative adversarial network The clothes generated by the 2020
clothing will take the latest fashion system is personalized for dif‑
SN Applied Sciences

trends and the clothes bought ferent types of consumers. This

by users as input and generate will help the manufacturing
new clothes recommendations. companies to come up with
the designs, which will directly
target the customer.
[49] Design and sale of sustainable Deep learning To develop an image-based Generative adversarial network The obtained system allows the 2022
clothing virtual fitting system that real‑ user to try on clothes from their
(2023) 5:387

istically reflects the appearance own home without having to

and the behavior of garment. go out to the stores. This pro‑
vides an advanced virtual shop‑
ping experience to the user.
[28] Design and sale of sustainable Deep learning To propose an intelligent, semi- Multi-layer feed-forward neural The proposed system allows 2020
clothing autonomous decision support network reducing long lead times,
system for the designers’ crea‑ eliminating the burden on
tive processes. the manufacturing process
and transforming traditional
processes into intelligent
processes.
[50] Design and sale of sustainable Machine learning To predict pattern and outfit Logistic regression The model analyzed colour, 2020
clothing based on the images collected pattern, and outfits and
from New York Fashion Week predicted the patterns that
Fall/Winter 2019 Instagram retailers might use in the com‑
posts. ing season for mass-market
consumers.
| https://doi.org/10.1007/s42452-023-05587-2
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year

[46] supply chain optimization Expert system Develop a framework to assess Fuzzy logic It was found that the important 2021
sustainability in multi-tier sup‑ indicators for assessing sustain‑
ply chains. ability are "Environmental
issues", "Economic issues", "Pol‑
SN Applied Sciences

icy and governance", "Participa‑

tion", "Social issues", "Trans‑
parency", "Leadership and
support". The expert system
allows a consistent assessment
of sustainability and provides
advice for decision making.
(2023) 5:387

[37] Design and sale of sustainable Expert system Help fashion designers in assess‑ Fuzzy logic The proposed process proved 2019
clothing ing consumers’ perception to be efficient when applied
of eco-style and ensure the to the analysis of the style of
success of sustainable product the garment with respect to
development. the eco-fashion style. This will
benefit fashion designers in
the design and development of
sustainable products.
[38] Reducing waste Expert system Development of a blockchain- Blockchain Three relevant issues were 2020
enabled circular supply chain identified for circular supply
management system in the fast chain management in fast
fashion industry with the goal fashion in terms of economic,
of reducing waste. environmental and social
responsibility. The proposed
blockchain-based system is
able to manage the challenges
in the fast-fashion supply
chains to achieve a zero-waste
circular economy.
[33] Supply chain optimization Deep learning To propose a model to detect Convolutional neural network The system showed good 2021
color differences in warp-knit‑ real-time performance and
ted fabric. accuracy and can meet the
| https://doi.org/10.1007/s42452-023-05587-2

fabric inspection requirements

of warp-knitted fabric factories.
[51] Design and sale of sustainable Deep learning To develop a clothing recom‑ Convolutional neural network The proposed model showed 2017
clothing mendation system that consid‑ high precision in recommend‑
ers the user’s social circle and ing personalized clothing.
fashion style consistency of However, it proved to be a bit
clothing items as important slow when exposed to massive
factors in the user’s decision data sets.
making.

Vol.:(0123456789)
Review
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year
Review

[24] Data analysis Machine learning To generate a better grouping K-means The proposal produces a smaller 2017
of products that have similar standard deviation in each

Vol:.(1234567890)
behavior in terms of sales to cluster compared to the con‑
provide better information for ventional K-means clustering
predictions, by a modification technique. This provides more
of the K-means algorithm. relevant and realistic informa‑
tion to use in predictions and
interpretations.
[47] Supply chain optimization Expert system Propose a system for the most Fuzzy logic The system showed its effective‑ 2021
appropriate sustainable selec‑ ness and applicability when
tion of suppliers in circular tested with a case study. Unlike
supply chains. other systems, the system
provides greater interactivity
with the user since it allows
customizing the evaluation
criteria and the importance of
SN Applied Sciences

the parameters.
[39] Reducing waste Deep learning Complement or replace manual Convolutional neural network The robot was successfully built 2021
waste sorting with an artificial and trained and installed near
intelligence-powered robot to Barcelona in a municipal waste
perform this task. sorting plant. It demonstrated
that this type of technology
is increasingly important in
(2023) 5:387

the transformation of supply

chains, from product design to
disposal.
[40] Reducing waste Deep learning Use different combinations of Convolutional neural network The databases created were 2019
cameras, lighting, and data used to train ANN systems
augmentation techniques to that obtained high accuracy
create image databases to train numbers in recognizing and
deep ANNs for recognition and classifying fiber materials. The
classification of fiber materials. effectiveness of data aug‑
mentation and ANNs applied
to waste classification was
evidenced.
[27] Supply chain optimization Expert system Development of an expert sys‑ Quality function deployment The system was tested with two 2018
tem that allows the identifica‑ study cases in which different
tion and selection of the best parameters and importance for
type of cotton fiber to create the different properties of the
products that meet the needs fibers were used. In both cases,
of customers. the system delivered a coher‑
ent result that, supported by
the numbers, was apt to satisfy
the user’s needs.
| https://doi.org/10.1007/s42452-023-05587-2
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year

[41] Reducing waste Deep learning Development of a methodology Convolutional neural network It was shown that using ANNs 2021
that allows the identification, can automatically identify and
classification and automatic classify different materials,
recycling of different materials. including glass, paper, plastic,
SN Applied Sciences

and organic matter. This and

similar systems could differenti‑
ate between materials that can
be cleaned for reuse and those
that can no longer be reused.
[42] Reducing waste Knowledge-based system Investigate the problems related Internet of Things The proposed system uses essen‑ 2020
to waste management and tial information on the truck,
(2023) 5:387

develop a system for the weight scale, collection and

sustainable and intelligent segregating room, conveyors,
management of these using hazardous bins, tumbler, heavy
industry 4.0 technologies. materials compartment, light
material compartment, and
final waste material bins. This
makes it possible to use all the
information and data on the
flow, input, process, and output
of waste for proper manage‑
ment of these.
[25] Supply chain optimization Deep learning Predicting the burst strength and Multi-layer feed-forward neural The ANN proved to be better 2012
air permeability of single jersey network than the linear regression and
knitted fabrics using regression allowed to find the critical
models and ANNs. factors influencing the air
permeability and the bursting
strength of the single jersey
fabrics with low error.
[26] Design and sale of sustainable Knowledge-based system Propose a color predicting Fuzzy logic The prediction system met the 2016
clothing method to capture the fashion trend color capture criteria as
trend, which is an essential fac‑ its predictions had low error
tor that leads to winning a sale. rates and high accuracy. The
| https://doi.org/10.1007/s42452-023-05587-2

system proved to have much

potential since it would allow
industries to make decisions to
select the color trend.
[30] Supply chain optimization Deep learning To propose a technique for Convolutional neural network The proposed model enhances 2020
tackling visual fashion clothes classification by effectively rec‑
analysis in images, aiming to ognizing key features and their
achieve clothing category clas‑ location in an image by repre‑
sification and attribute predic‑ senting multilevel contextual
tion by producing regularized information of landmarks.
landmark layouts.

Vol.:(0123456789)
Review
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year
Review

[53] Design and sale of sustainable Deep learning To propose a neural network Generative adversarial network The proposed system generates 2020
clothing system to design and recom‑ attractive designs appreciated

Vol:.(1234567890)
mend new clothing patterns by users. This was evidenced in
and styles. the evaluations since the sys‑
tem was evaluated with online
surveys and the volunteers
prefer the clothing generated
by the proposal.
[36] Supply chain optimization Machine learning Address the problem of fabric Support vector machine The results showed that the use 2011
defect recognition from images of support vector machine
using ML techniques. for the recognition of tissue
defects produces accurate
results, and it emerged that
this type of classifier has great
potential for the automatic
inspection of tissue defects in
SN Applied Sciences

the industry.
[52] Supply chain optimization Machine learning To define a methodology to K-means The porposed technique facili‑ 2017
obtain a clustering of human tates the automatic handling of
morphology shapes repre‑ 3D scans and calculating their
sentative of a population and geodesic shape description
to extract the most significant with optimized performance
morphotype of each class. for maximum differentiation
(2023) 5:387

ability. This approach has

the potential to significantly
enhance the sizing systems
used by clothing companies.
[32] Supply chain optimization Deep learning To use ANNs and regression Multi-layer feed-forward neural The use of an ANN provided the 2010
methods for exploring the rela‑ network most comprehensive quan‑
tionship between various drape titative understanding of a
parameters and mechanical material’s drapability, includ‑
fabric properties. ing shape parameters and
measurement technology. This
makes it highly beneficial for
producing high-quality textiles
and garments.
[44] Reducing waste Deep learning To design a time series model for Multi-layer feed-forward neural The proposed model showed to 2018
predicting monthly based solid network give accurate predictive results.
waste generation. The error percentages obtained
were low and allowed to obtain
a realistic prediction about the
generation of garbage.
| https://doi.org/10.1007/s42452-023-05587-2
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year

[45] Reducing waste Deep learning To develop a model to estimate Multi-layer feed-forward neural Different approaches were 2019
the annual amount of packag‑ network examined, and the ANN was
ing waste. the one that produced the best
results. Compared to conven‑
SN Applied Sciences

tional regression procedures,

the ANN was demonstrated to
have better explanatory power
and to be less susceptible to
outliers. The technique can be
used in other areas like fashion,
manufacturing, and construc‑
tion because the forecasts were
(2023) 5:387

accurate.
[58] Supply chain optimization Deep learning To test whether a simplified Multi-layer feed-forward neural The results showed that the ANN 2016
neural-network computational network model performs better than
model can make routing deci‑ the best-tested routing heuris‑
sions in a logistics facility more tics. Simulations with complex
efficiently than five intelligent logistics scenarios were carried
routing heuristics. out with real-world data, and
the ANN showed potential to
be applied in other fields, such
as supply chains and public
transport.
[29] Data analysis Deep learning Propose a system for fast and Multi-layer feed-forward neural It was found that using ML and 2011
efficient sales forecasting for network traditional statistical methods
fashion products. allows for controlling all the
time cost and forecast errors
so that a fast and effective
forecast can be obtained. The
system made sales forecasts
accurately.
[54] Design and sale of sustainable Expert system To propose an hybrid model of Multi-layer feed-forward neural The system proposed demon‑ 2013
clothing intelligent classification for size network strated high accuracy and can
| https://doi.org/10.1007/s42452-023-05587-2

recommendation. serve as a size recommenda‑

tion expert, reducing time and
increasing customer satisfac‑
tion by identifying the ideal
clothing size.

Vol.:(0123456789)
Review
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year
Review

[43] Supply chain optimization Deep learning To examine the effects of waste Multi-layer feed-forward neural According to the proposed 2019
material composition on the network model, investigations on the

Vol:.(1234567890)
best truck route’s time, dis‑ optimization of geographic
tance, and air emissions. Create information system routes
a model that predicts the rate must take into account
of garbage generation, and temporal variations in trash
identify optimal waste collec‑ composition and characteris‑
tion routes with minimal travel tics. Additionally, the model’s
distance. error was moderately low and
allowed the optimization of the
routes.
[23] Supply chain optimization Expert system To propose a decision support Fuzzy logic The system demonstrated its 2013
model for determining the ability to supply crucial data
chance of meeting on-time to the production and supply
delivery in a complex supply chain planning departments
chain environment. within the organization, while
SN Applied Sciences

also offering real-time monitor‑

ing of delivery timeliness.
[22] Supply chain optimization Deep learning To propose a system to deter‑ Multi-layer feed-forward neural The model was successful in 2011
mine the optimum level of network terms of its agreement with
finished goods inventory. actual values in a manufac‑
turing industry, with a good
performance by the ANN and
(2023) 5:387

a low mean absolute error

percentage.
[55] Design and sale of sustainable Deep learning To propose a novel ANN frame‑ Convolutional neural network The experiments showed that 2019
clothing work that simultaneously the proposal significantly
provides outfit recommenda‑ improved over state-of-the-art
tions and generates abstractive baselines for outfit recommen‑
comments. dation. In addition to generat‑
ing the recommendation, the
system generates explanations
for these results with high
precision compared to human-
written comments.
[56] Design and sale of sustainable Deep learning To propose an ANN model to Generative adversarial network The proposed method can effec‑ 2020
clothing generate complementary tively assess the compatibility
fashion items by measuring of complementary fashion
the compatibility between items both on an item-to-item
complementary fashion items basis and in item-template.
comprehensively from both Results of testing indicated that
item-item and item-template the proposed scheme has a
perspectives. high level of accuracy.
| https://doi.org/10.1007/s42452-023-05587-2
 Table 8  (continued)
References Domain AI class Aim Main technique Key findings Year

[34] Data analysis Machine learning To apply image mining for Mean-shift clustering The suggested method dem‑ 2020
fashion analysis, clustering onstrated its robustness in
and predicting using fashion- handling uncertainty present in
related images collected from the input images. Additionally,
SN Applied Sciences

the social network. it proved to be beneficial for

the fashion industry in making
informed decisions regarding
the production of fashion items
with specific features.
[31] Supply chain optimization Machine learning To use data mining and symme‑ Decision tree, naive Bayes, ran‑ The study found that the random 2020
try-based learning techniques dom forest, and Bayesian forest forest classifier performed
(2023) 5:387

on product data to create a better than other methods in

classification model for predict‑ classifying garments and sub‑
ing the garment category and categories (upper body, lower
for predicting the garment body, whole-body). These
sub-category. classifications are useful for
decision support systems and
recommendation algorithms,
aiding customers in making
informed purchases.
| https://doi.org/10.1007/s42452-023-05587-2

Vol.:(0123456789)
Review
Review SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

long-term costs and waste generation. Promising results in evaluating their operations and making decisions to
were obtained through real case testing. achieve sustainability objectives.
Likewise, in Ref. [25], ANNs are employed to discover Common limitations in these articles include the high
the properties influencing burst strength and air per‑ data and processing demands of some AI approaches,
meability in single jersey knitted fabrics, as well as to such as those based on ANNs. Additionally, the reluctance
predict these properties. In [32], ANNs are used to pre‑ to adopt this technology due to a lack of knowledge can
dict the relationship between drape parameters and hinder its widespread use. Nonetheless, these works col‑
fabric mechanical properties. These AI applications pro‑ lectively demonstrate AI’s substantial potential to optimize
vide quantitative insights into material characteristics various aspects of a fashion company.
and their influencing parameters, facilitating resource
optimization and high-quality garment and textile 4.2 Design and sale of sustainable clothing
manufacturing.
In the first case, it was found that the burst strength of Sustainable fashion is often perceived as less exciting, of
single knit fabrics is affected by fiber strength, fiber elon‑ lower quality, or not aligned with the latest fashion trends
gation, and fiber mean length. Similarly, air permeability [63]. Despite this misconception, the sustainable fashion
is affected by fiber mean length, yarn twists per inch, yarn industry and its products often struggle to gain market rel‑
count and number of wales and courses. In the second evance [64]. However, AI emerges as a potentially valuable
case, bending, shear and aerial density were found to tool for sustainable fashion companies, aiding in both the
affect the drape parameters the most. In both cases, AI design and promotion of clothing that aligns with current
helped to understand quantitatively and in more detail fashion trends [65]. These technologies can significantly
the materials’ characteristics and the parameters that influ‑ contribute to driving the purchase and use of sustainable
ence them. garments [66].
In a complementary vein [22], employs ANNs to deter‑ For instance, in Ref. [28], an application of Artificial Neu‑
mine the optimal inventory level for finished products, ral Networks (ANNs) is proposed to create a semi-autono‑
considering setup costs, holding costs, material costs, and mous intelligent system supporting designers during the
product demand. This optimization prevents overproduc‑ creative process. This system leverages user preferences,
tion of merchandise that may go unsold, ultimately reduc‑ fashion trends, seasonal data, and company constraints to
ing waste and resource expenditure. make predictions and design suggestions.
In terms of resource and product distribution, AI also Similarly, in Ref. [37], an expert system is employed to
proves beneficial. For instance [58, 43], utilize ANNs to assess consumer perceptions of eco-style. The objective
address routing problems, optimizing distribution vehi‑ is to analyze consumers and gain deeper insights into
cle routes to minimize distances, reduce gas emissions, eco-fashion and consumer perceptions, thereby ensur‑
and ensure on-time product delivery based on real-time ing the success of eco-fashion and sustainable product
geographical context information. development.
Furthermore, in Ref. [23], a decision support model lev‑ Other works focus on analyzing the latest fashion trends
erages fuzzy logic to predict on-time delivery chances in to inform clothing design. In Ref. [50], machine learning
a complex supply chain environment, mitigating negative methods are used to analyze trends from events like the
consequences of delivery variations, demand forecasting New York Fashion Week, predicting new design patterns
inaccuracies, materials shortages, and distribution lead based on this data. Likewise, in Ref. [26], a fuzzy logic-
time uncertainties. based system is proposed to analyze fashion trends related
Addressing supplier selection in sustainable supply to color and suggest new color combinations for manufac‑
chain management [47], presents an expert system for turers to consider. Both approaches have been evaluated
circular supply chains that manufacture, dispose of, and and demonstrated their utility in supporting designers’
recycle, reducing costs and waste. The system combines decision-making processes.
multi-criteria decision-making, ML, and fuzzy logic to Design suggestion systems are also explored in Refs.
select the most suitable suppliers, as demonstrated in a [48, 53], but with a twist. These works introduce Genera‑
real-world case study. tive Adversarial Networks (GANs) to generate new clothing
Finally, [46] introduces an expert system based on fuzzy designs based on fashion trends and user purchasing data.
logic to evaluate supply chain sustainability comprehen‑ What sets these systems apart is their ability to generate
sively. This system analyzes various aspects, including envi‑ recommendations and graphical design suggestions. This
ronmental, economic, policy, governance, participation, innovation streamlines the design and manufacturing pro‑
social issues, transparency, and leadership support, yield‑ cess, offering designers textual feedback and visual repre‑
ing a sustainability score that assists fashion companies sentations, significantly enhancing efficiency.

Vol:.(1234567890)
SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2 Review

While these developments are instrumental in making development, including well-being, health, clean water,
sustainable fashion more appealing and aligned with cur‑ and climate change.
rent fashion trends, the ultimate goal of gaining market Within the realm of supply chains [38], introduces an
relevance hinges on increasing the attractiveness and expert system aimed at waste management. This system
accessibility of sustainable clothing for consumers. This focuses on the return of products at the end of their life
can be achieved through technologies that promote sus‑ cycle to various supply chain components for reuse and
tainable clothing and improve the search and purchasing value recovery. Notably, it incorporates blockchain tech‑
experience. nology, enabling the transparent processing and record‑
For instance, in Ref. [51], CNNs are employed to provide ing of data across the entire product lifecycle, thus foster‑
recommendations that consider not only customer pref‑ ing a circular economy.
erences but also their social network. A similar approach AI’s predictive capabilities are harnessed in waste pre‑
is proposed in [55], which generates visual recommenda‑ diction models like those seen in Refs. [44, 45], both uti‑
tions along with explanations for the recommendations. lizing ANNs. By leveraging historical data, these systems
In Ref. [56], GANs are used to recommend complementary provide more accurate waste generation predictions.
fashion items, assisting customers in completing their out‑ Consequently, these insights empower the development
fits by suggesting items that complement their selections. of strategies to curtail waste, boost recycling rates, and
Furthermore, in Ref. [54], a system is introduced to offer promote sustainability.
size recommendations, utilizing size tables and ANNs to AI-driven waste classification is a rapidly advancing
create an intelligent sizing system. This system was tested field. For instance [40], presents a methodology reliant on
in an Iranian store, resulting in time savings and increased CNNs for classifying different fiber materials, even when
customer satisfaction by assisting customers in selecting confronted with limited data. Similarly Ref. [41], demon‑
the right clothing size. strates the effectiveness of CNNs in classifying various
Continuing to enhance the shopping experience, a vir‑ materials, including glass, paper, plastic, and organic mat‑
tual try-on interface based on GANs is presented in Ref. ter, using images generated from smartphones.
[49]. This interface allows users to virtually try on clothes Taking waste sorting to the next level [39], combines
from the comfort of their homes, facilitating online shop‑ CNNs with robotic technology. Equipped with sensors
ping and providing a realistic visualization of how the and mechanical grippers, a robot continuously monitors
clothes would appear when worn. waste flow and autonomously performs sorting tasks.
In conclusion, AI has the potential to reshape sustain‑ The practical deployment of this system in a major Span‑
able fashion by aligning it with current trends, enhancing ish waste sorting plant underscores its industry relevance
the design process, and improving the shopping experi‑ and potential for efficient waste management.
ence. This technology bridges the gap between sustain‑ In summary, AI presents a powerful toolkit for address‑
ability and market relevance, making eco-friendly fashion ing the fashion industry’s waste management challenges.
more appealing and accessible to consumers. These innovative applications not only offer solutions for
sustainable waste disposal but also align with broader
4.3 Reducing waste sustainability goals. By leveraging AI for waste prediction,
classification, and intelligent systems, fashion companies
Effective waste management is critical for the fashion can not only reduce their environmental impact but also
industry’s sustainability, especially considering its histori‑ enhance their operational efficiency. These advancements
cal negative environmental impact [67]. Over recent years, underscore the transformative role of AI in promoting sus‑
the proliferation of fast fashion and a throwaway culture tainability within the fashion industry, emphasizing the
has led to a significant surge in textile production and path towards a more eco-friendly and responsible future.
consumption [68]. Unfortunately, the majority of textiles
and clothing ultimately find their way into landfills, with 4.4 Data analysis
only a small fraction being recycled, making textile waste
a pressing global concern [69]. Data analysis plays a pivotal role in the fashion industry,
AI offers innovative solutions to address these waste offering valuable insights and predictive capabilities that
management challenges. In Ref. [42], an intelligent knowl‑ can be harnessed to drive sustainability efforts [57, 70, 71].
edge-based system is applied to sustainable waste man‑ Customer segmentation, as demonstrated in studies like
agement. This comprehensive approach analyzes various [21, 24], allows fashion companies to gain a deeper under‑
facets of waste collection, transportation, and processing. standing of their customer base. By tailoring strategies to
Moreover, it considers critical dimensions of sustainable specific market segments, these companies can effectively

Vol.:(0123456789)
Review SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

promote and sell sustainable clothing to a diverse range of systematic review of 37 articles sourced from relevant
consumers. online publications. These articles were subsequently
Moreover, predictive analytics, exemplified in research reviewed and analyzed.
such as Refs. [29, 34], empowers fashion businesses to make The analysis of the selected articles unequivocally dem‑
informed decisions and anticipate market trends. Sales pre‑ onstrates that AI plays a pivotal role in the fashion indus‑
dictions and style forecasting enable these companies to try’s transition towards sustainable development. The
optimize inventory management, streamline supply chains, findings underscore a multitude of contributions that can
and proactively address future challenges, all of which are be harnessed across various facets of the fashion indus‑
essential for sustainable fashion practices. try, encompassing supply chain optimization, sustainable
In the context of sustainable fashion, studies like [35] shed clothing design and sales, waste management and control,
light on consumer behavior and attitudes towards sustain‑ and data analysis.
ability. By utilizing ML techniques to analyze consumer Our analysis reveals the pervasive presence of ANNs as
responses, these studies reveal valuable insights into con‑ the primary technological cornerstone in the integration
sumer knowledge and preferences regarding sustainable of AI within the fashion industry. ANNs, with their versatile
clothing. The findings underscore the importance of edu‑ capabilities, occupy a central role across a diverse array
cating consumers about sustainable fashion practices and of applications. These encompass but are not limited to
strategies to encourage their adoption. waste classification, where ANNs excel in their ability to
In sum, data analysis serves as a powerful tool for fash‑ accurately categorize materials. Furthermore, they play a
ion companies striving to enhance sustainability. From pivotal role in garment defect detection, swiftly identify‑
customer segmentation to predictive analytics, these ing and rectifying manufacturing flaws. In addition, ANNs
data-driven approaches empower the fashion industry to significantly contribute to the augmentation of sustain‑
make informed decisions and tailor strategies that promote able clothing design and production, ensuring that the
sustainable clothing consumption and production. This industry’s environmental footprint is minimized.
underscores the potential of data analysis in driving posi‑ Moreover, our investigation uncovered a balanced dis‑
tive change within the fashion industry. tribution among sustainability dimensions in the reviewed
articles. The majority of contributions are situated within
4.5 Limitations of the study the economic dimension, prioritizing the fulfillment of
current and future customer needs. Subsequently, the
One notable limitation of this study is its primary reliance environmental dimension encompasses a significant por‑
on existing literature and readily available sources. This tion of articles, primarily focusing on responsible waste
approach may inadvertently omit certain non-academic, management and the promotion of recycling. Finally,
ongoing, or unpublished works that could potentially con‑ the social dimension concentrates on creating accessible
tribute valuable insights to the field of AI in sustainable fash‑ environments and providing services that cater to diverse
ion. Nevertheless, it is worth noting that a comprehensive market needs.
array of relevant online sources was diligently incorporated, The articles predominantly targeted the business and
making a concerted effort to ensure that the findings pre‑ customer sectors. In the business sector, contributions
sented in this paper offer a representative and meaningful primarily centered on improving garment manufacturing,
overview. design, and handling processes. Conversely, the customer
Furthermore, this study primarily operates at a high-level sector emphasized equipping users with tools to encour‑
examination of the contributions of AI within the sustainable age the purchase of sustainable products while ensuring
fashion industry. This means that it provides a broad over‑ a delightful and attractive shopping experience. Notably,
view and general insights into how AI is being applied to there were no significant contributions directed towards
promote sustainability in the fashion sector. However, it may the government sector.
not encompass the intricate details and specific nuances However, it’s crucial to acknowledge the limitations
that some readers, particularly those looking for in-depth encountered on this journey towards AI-driven sustainabil‑
technical or sector-specific information, may seek. ity in fashion. One notable challenge lies in the substantial
volume of data required to train neural networks effec‑
tively. Acquiring and managing extensive datasets, espe‑
5 Conclusions cially in a domain as dynamic as fashion, can prove daunt‑
ing. The fashion industry also faces the cost considerations
This study aimed to assess the application of Artificial associated with the implementation of AI technologies.
Intelligence (AI) in the fashion industry to promote These investments encompass not only the acquisition of
sustainability. To achieve this goal, we conducted a cutting-edge hardware and software but also the training

Vol:.(1234567890)
SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2 Review

and upskilling of personnel to harness the full potential of source, provide a link to the Creative Commons licence, and indicate
AI systems. if changes were made. The images or other third party material in this
article are included in the article’s Creative Commons licence, unless
indicated otherwise in a credit line to the material. If material is not
included in the article’s Creative Commons licence and your intended
6 Future work use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright
holder. To view a copy of this licence, visit http://​creat​iveco​mmons.​
As of 2023, the landscape of AI has undergone a remark‑ org/​licen​ses/​by/4.​0/.
able surge in popularity, marking a global trend like never
before. This surge has permeated various industries,
including fashion, where AI’s transformative potential
is becoming increasingly evident. Therefore, for future References
research endeavors, it is highly advisable to conduct a
dedicated review with a specific focus on the develop‑ 1. Hawken P, Lovins AB, Lovins LH (1999) Natural capitalism: creat‑
ing the next industrial revolution. Little Brown, Boston
ments and impacts witnessed in this pivotal year. 2. Moretto A, Macchion L, Lion A, Caniato F, Danese P, Vinelli A
An exploration centered around the year 2023 can shed (2018) Designing a roadmap towards a sustainable supply
light on how this unprecedented surge in AI’s popularity chain: a focus on the fashion industry. J Clean Prod 193:169–184.
has resonated within the fashion industry. This examina‑ https://​doi.​org/​10.​1016/J.​JCLEP​RO.​2018.​04.​273
3. Jamwal A, Agrawal R, Sharma M, Kumar V, Kumar S (2021) Devel‑
tion can provide valuable insights into how the fashion oping a sustainability framework for industry 4.0. Procedia CIRP
sector has harnessed the momentum of AI, whether it 98:430–435. https://​doi.​org/​10.​1016/J.​PROCIR.​2021.​01.​129
be through innovative applications, novel solutions, or 4. Ruggerio CA (2021) Sustainability and sustainable develop‑
heightened integration. Such a study can also help identify ment: a review of principles and definitions. Sci Total Environ
786:147481. https://​doi.​org/​10.​1016/J.​SCITO​TENV.​2021.​147481
emerging trends, challenges, and opportunities unique to 5. Kristensen HS, Mosgaard MA (2020) A review of micro level
this period, offering a comprehensive view of the indus‑ indicators for a circular economy—moving away from the three
try’s trajectory. dimensions of sustainability? J Clean Prod 243:118531. https://​
Moreover, a complementary study centered around doi.​org/​10.​1016/j.​jclep​ro.​2019.​118531
6. Becker C, Betz S, Chitchyan R, Duboc L, Easterbrook SM, Pen‑
2023 would not only serve as a testament to the dyna‑ zenstadler B, Seyff N, Venters CC (2016) Requirements: the key
mism of AI but also provide an invaluable perspective to sustainability. IEEE Softw 33:56–65. https://​doi.​org/​10.​1109/​
for researchers, businesses, and stakeholders seeking to MS.​2015.​158
navigate the evolving landscape of AI-driven sustainabil‑ 7. Alshehhi A, Nobanee H, Khare N (2018) The impact of sustain‑
ability practices on corporate financial performance: literature
ity in fashion. It can illuminate how the fashion industry trends and future research potential. Sustainability. https://​doi.​
has adapted and innovated in response to the surge in org/​10.​3390/​su100​20494
AI adoption, potentially uncovering novel strategies and 8. Huang L, Wu J, Yan L (2015) Defining and measuring urban sus‑
best practices for sustainable growth and development. tainability: a review of indicators. Landscape Ecol 30:1175–1193.
https://​doi.​org/​10.​1007/​S10980-​015-​0208-2/​METRI​CS
9. Carlsen L, Bruggemann R (2022) The 17 united nations’ sustain‑
able development goals: a status by 2020. Int J Sustain Dev
Author Contributions LR: Conceptualization, Formal analysis and World Ecol 29(3):219–229. https://​doi.​org/​10.​1080/​13504​509.​
investigation, Writing—original draft preparation, Writing—review 2021.​19484​56
and editing. FRE, AGP, and EC: Formal analysis and investigation, Writ‑ 10. Wren B (2022) Sustainable supply chain management in the fast
ing—review and editing. All authors reviewed the manuscript. fashion industry: a comparative study of current efforts and best
practices to address the climate crisis. Clean Logistics Supply
Funding No funding was received to assist with the preparation of Chain 4:100032. https://​doi.​org/​10.​1016/J.​CLSCN.​2022.​100032
this manuscript. 11. Akter MMK, Haq UN, Islam MM, Uddin MA (2022) Textile-apparel
manufacturing and material waste management in the circular
Data availability Data sharing not applicable to this article as no economy: a conceptual model to achieve sustainable devel‑
datasets were generated or analysed during the current study. opment goal (SDG) 12 for Bangladesh. Clean Environ Syst
4:100070. https://​doi.​org/​10.​1016/J.​CESYS.​2022.​100070
12. Hossain MN, Rahman MM, Afrin S, Akbor MA, Siddique MAB,
Declarations Malafaia G (2023) Identification and quantification of microplas‑
tics in agricultural farmland soil and textile sludge in Bangla‑
Conflicts of interest The authors have no relevant financial interests desh. Sci Total Environ 858:160118. https://​doi.​org/​10.​1016/J.​
in the manuscript and no other potential conflicts of interest to dis‑ SCITO​TENV.​2022.​160118
close. 13. Hossain L, Sarker SK, Khan MS (2018) Evaluation of present
and future wastewater impacts of textile dyeing industries in
Open Access This article is licensed under a Creative Commons Attri‑ Bangladesh. Environ Dev 26:23–33. https://​doi.​org/​10.​1016/J.​
bution 4.0 International License, which permits use, sharing, adap‑ ENVDEV.​2018.​03.​005
tation, distribution and reproduction in any medium or format, as 14. Bruzzone AG, Massei M, Frosolini M (2022) Redesign of sup‑
long as you give appropriate credit to the original author(s) and the ply chain in fashion industry based on strategic engineering.

Vol.:(0123456789)
Review SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2

Procedia Comput Sci 200:1913–1918. https://​d oi.​o rg/​1 0.​ 29. Yu Y, Choi T-M, Hui C-L (2011) An intelligent fast sales forecasting
1016/J.​PROCS.​2022.​01.​392 model for fashion products. Expert Syst Appl 38(6):7373–7379.
15. Ramos L (2022) Artificial intelligence for cancer detection https://​doi.​org/​10.​1016/j.​eswa.​2010.​12.​089
using medical image: highlights and limitations. Green World 30. Shajini M, Ramanan A (2021) An improved landmark-driven
J. https://​doi.​org/​10.​53313/​gwj51​011 and spatial-channel attentive convolutional neural network for
16. Pournader M, Ghaderi H, Hassanzadegan A, Fahimnia B (2021) fashion clothes classification. Visual Computer 37:1517–1526.
Artificial intelligence applications in supply chain manage‑ https://​doi.​org/​10.​1007/​S00371-​020-​01885-7/​METRI​CS
ment. Int J Prod Econ 241:108250. https://​doi.​org/​10.​1016/J.​ 31. Jain S, Kumar V(2020) Garment categorization using data mining
IJPE.​2021.​108250 techniques. Symmetry 12(6) . https://​doi.​org/​10.​3390/​sym12​
17. Thomassey S, Zeng X (2018) Artificial intelligence for fashion 060984
industry in the big data era. Springer Ser Fashion Bus. https://​ 32. Pattanayak AK, Luximon A, Khandual A (2010) Prediction of
doi.​org/​10.​1007/​978-​981-​13-​0080-6_1 drape profile of cotton woven fabrics using artificial neural net‑
18. Ahmad T, Zhang D, Huang C, Zhang H, Dai N, Song Y, Chen work and multiple regression method. Text Res J 81:559–566.
H (2021) Artificial intelligence in sustainable energy indus‑ https://​doi.​org/​10.​1177/​00405​17510​380783
try: status quo, challenges and opportunities. J Clean Prod 33. Guosheng X, Yang X, Zhiqi Y, Yize S (2022) An intelligent defect
289:125834. https://​doi.​org/​10.​1016/J.​JCLEP​RO.​2021.​125834 detection system for warp-knitted fabric. Text Res J 92(9–
19. ...Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann 10):1394–1404. https://​doi.​org/​10.​1177/​00405​17521​10600​84
TC, Mulrow CD, Shamseer L, Tetzlaff JM, Akl EA, Brennan SE, 34. Wazarkar S, Keshavamurthy BN (2020) Social image mining for
Chou R, Glanville J, Grimshaw JM, Hróbjartsson A, Lalu MM, fashion analysis and forecasting. Appl Soft Comput 95:106517.
Li T, Loder EW, Mayo-Wilson E, McDonald S, McGuinness LA, https://​doi.​org/​10.​1016/j.​asoc.​2020.​106517
Stewart LA, Thomas J, Tricco AC, Welch VA, Whiting P, Moher D 35. Garcia S, Cordeiro A, de Alencar Nääs I, de Oliveira Costa Neto
(2021) The Prisma 2020 statement: an updated guideline for P.L (2019) The sustainability awareness of Brazilian consumers
reporting systematic reviews. BMJ. https://​doi.​org/​10.​1136/​ of cotton clothing. J Clean Prod 215:1490–1502. https://d ​ oi.o​ rg/​
BMJ.​N71 10.​1016/j.​jclep​ro.​2019.​01.​069
20. Rethlefsen, M.L., Kirtley, S., Waffenschmidt, S., Ayala, A.P., 36. Ghosh A, Guha T, Bhar RB, Das S (2011) Pattern classification
Moher, D., Page, M.J., Koffel, J.B., Blunt, H., Brigham, T., Chang, of fabric defects using support vector machines. Int J Cloth Sci
S., Clark, J., Conway, A., Couban, R., de Kock, S., Farrah, K., Technol 23:142–151. https://d ​ oi.o
​ rg/1
​ 0.1
​ 108/0 ​ 95562 ​ 21111 ​ 1073​
Fehrmann, P., Foster, M., Fowler, S.A., Glanville, J., Harris, E., 33
Hoffecker, L., Isojarvi, J., Kaunelis, D., Ket, H., Levay, P., Lyon, 37. Wagner M, Curteza A, Hong Y, Chen Y, Thomassey S, Zeng X
J., McGowan, J., Murad, M.H., Nicholson, J., Pannabecker, V., (2019) A design analysis for eco-fashion style using sensory
Paynter, R., Pinotti, R., Ross-White, A., Sampson, M., Shields, evaluation tools: consumer perceptions of product appearance.
T., Stevens, A., Sutton, A., Weinfurter, E., Wright, K., Young, S., J Retail Consum Serv 51:253–262. https://d ​ oi.o​ rg/1
​ 0.1​ 016/j.j​ retc​
Group, P.-S. (2021) Prisma-s: an extension to the Prisma state‑ onser.​2019.​06.​005
ment for reporting literature searches in systematic reviews. 38. Wang B, Luo W, Zhang A, Tian Z, Li Z (2020) Blockchain-enabled
Syst Rev 10:39. https://​doi.​org/​10.​1186/​s13643-​020-​01542-z circular supply chain management: a system architecture for
21. Brito PQ, Soares C, Almeida S, Monte A, Byvoet M (2015) Cus‑ fast fashion. Comput Ind 123:103324. https://​doi.​org/​10.​1016/j.​
tomer segmentation in a large database of an online custom‑ compi​nd.​2020.​103324
ized fashion business. Robot Comput 36:93–100. https://​doi.​ 39. Wilts H, Garcia BR, Garlito RG, Gómez LS, Prieto EG (2021) Artifi‑
org/​10.​1016/j.​rcim.​2014.​12.​014 cial intelligence in the sorting of municipal waste as an enabler
22. Paul SK, Azeem A (2011) An artificial neural network model of the circular economy. Resources. https://​doi.​org/​10.​3390/​
for optimization of finished goods inventory. Int J Ind Eng resou​rces1​00400​28
Comput 2:431–438. https://​doi.​org/​10.​5267/j.​ijiec.​2011.​01.​005 40. Vrancken C, Longhurst P, Wagland S (2019) Deep learning in
23. Nakandala D, Samaranayake P, Lau HCW (2013) A fuzzy-based material recovery: development of method to create training
decision support model for monitoring on-time delivery database. Expert Syst Appl 125:268–280. https://​doi.​org/​10.​
performance: a textile industry case study. Eur J Oper Res 1016/j.​eswa.​2019.​01.​077
225(3):507–517. https://​doi.​org/​10.​1016/j.​ejor.​2012.​10.​010 41. Nañez Alonso SL, Reier Forradellas RF, Pi Morell O, Jorge-Vazquez
24. Tehrani AF, Ahrens D (2017) Modified sequential k-means clus‑ J (2021) Digitalization, circular economy and environmental sus‑
tering by utilizing response: a case study for fashion products. tainability: the application of artificial intelligence in the effi‑
Expert Syst 34:12226. https://​doi.​org/​10.​1111/​EXSY.​12226 cient self-management of waste. Sustainability. https://​doi.​org/​
25. Unal PG, Üreyen ME, Mecit D (2012) Predicting properties of 10.​3390/​su130​42092
single jersey fabrics using regression and artificial neural net‑ 42. Fatimah YA, Govindan K, Murniningsih R, Setiawan A (2020)
work models. Fibers Polym 13:87–95. https://​doi.​org/​10.​1007/​ Industry 4.0 based sustainable circular economy approach
S12221-​012-​0087-Y/​METRI​CS for smart waste management system to achieve sustainable
26. Hsiao SW, Lee CH, Chen RQ, Yen CH (2017) An intelligent sys‑ development goals: a case study of Indonesia. J Clean Prod
tem for fashion colour prediction based on fuzzy c-means 269:122263. https://​doi.​org/​10.​1016/j.​jclep​ro.​2020.​122263
and gray theory. Color Res App 42:273–285. https://​doi.​org/​ 43. Vu HL, Bolingbroke D, Ng KTW, Fallah B (2019) Assessment of
10.​1002/​COL.​22057 waste characteristics and their impact on GIS vehicle collection
27. Chakraborty S, Prasad K (2018) A quality function deployment- route optimization using ANN waste forecasts. Waste Manage
based expert system for cotton fibre selection. J Institut Eng 88:118–130. https://​doi.​org/​10.​1016/j.​wasman.​2019.​03.​037
99:43–53. https://​doi.​org/​10.​1007/​S40034-​018-​0111-X/​METRI​ 44. Singh D, Satija A (2018) Prediction of municipal solid waste
CS generation for optimum planning and management with arti‑
28. Papachristou E, Chrysopoulos A, Bilalis N (2021) Machine learn‑ ficial neural network-case study: Faridabad city in Haryana state
ing for clothing manufacture as a mean to respond quicker and (India). Int J Syst Assurance Eng Managem 9:91–97. https://​doi.​
better to the demands of clothing brands: a greek case study. org/​10.​1007/​S13198-​016-​0484-5/​METRI​CS
Int J Adv Manuf Technol 115:691–702. https://​doi.​org/​10.​1007/​ 45. Oliveira V, Sousa V, Dias-Ferreira C (2019) Artificial neural
S00170-​020-​06157-1/​METRI​CS network modelling of the amount of separately-collected

Vol:.(1234567890)
SN Applied Sciences (2023) 5:387 | https://doi.org/10.1007/s42452-023-05587-2 Review

household packaging waste. J Clean Prod 210:401–409. https://​ 59. Carter CR, Rogers DS, Choi TY (2015) Toward the theory of the
doi.​org/​10.​1016/j.​jclep​ro.​2018.​11.​063 supply chain. J Supply Chain Manag 51(2):89–97. https://​doi.​
46. Shayganmehr M, Kumar A, Luthra S, Garza-Reyes JA (2021) org/​10.​1111/​jscm.​12073
A framework for assessing sustainability in multi-tier supply 60. Wong C, Skipworth H, Godsell J, Achimugu N (2012) Towards a
chains using empirical evidence and fuzzy expert system. J theory of supply chain alignment enablers: a systematic litera‑
Clean Prod 317:128302. https://​doi.​org/​10.​1016/j.​jclep​ro.​2021.​ ture review. Supply Chain Manag 17:419–437. https://​doi.​org/​
128302 10.​1108/​13598​54121​12465​67/​FULL/​XML
47. Alavi B, Tavana M, Mina H (2021) A dynamic decision support 61. Farooque M, Zhang A, Thürer M, Qu T, Huisingh D (2019) Circular
system for sustainable supplier selection in circular economy. supply chain management: a definition and structured litera‑
Sustain Prod Consumpt 27:905–920. https://​doi.​org/​10.​1016/j.​ ture review. J Clean Prod 228:882–900. https://d ​ oi.o
​ rg/1 ​ 0.1
​ 016/j.​
spc.​2021.​02.​015 jclep​ro.​2019.​04.​303
48. Singh M, Bajpai U, Vijayarajan V, Prasath S (2020) Generation 62. Sharma R, Shishodia A, Gunasekaran A, Min H, Munim ZH (2022)
of fashionable clothes using generative adversarial networks. The role of artificial intelligence in supply chain management:
Int J Cloth Sci Technol 32:177–187. https://​doi.​org/​10.​1108/​ mapping the territory. Int J Prod Res 60(24):7527–7550. https://​
IJCST-​12-​2018-​0148 doi.​org/​10.​1080/​00207​543.​2022.​20296​11
49. Ghodhbani H, Neji M, Qahtani AM, Almutiry O, Dhahri H, Alimi 63. Dabas CS, Whang C (2022) A systematic review of drivers of sus‑
AM (2022) Dress-up: deep neural framework for image-based tainable fashion consumption: 25 years of research evolution. J
human appearance transfer. Multimedia Tools App. https://​doi.​ Glob Fash Market 13(2):151–167. https://d ​ oi.o
​ rg/1
​ 0.1​ 080/2 ​ 0932​
org/​10.​1007/​S11042-​022-​14127-W/​FIGUR​ES/​16 685.​2021.​20160​63
50. Chakraborty S, Hoque SMA, Kabir SMF (2020) Predicting fashion 64. Hur E, Cassidy T (2019) Perceptions and attitudes towards sus‑
trend using runway images: application of logistic regression in tainable fashion design: challenges and opportunities for imple‑
trend forecasting. Int J Fashion Design Tech Edu 13(3):376–386. menting sustainability in fashion. Int J Fashion Design Technol
https://​doi.​org/​10.​1080/​17543​266.​2020.​18290​96 Educ 12(2):208–217. https://​doi.​org/​10.​1080/​17543​266.​2019.​
51. Sun GL, Cheng ZQ, Wu X, Peng Q (2018) Personalized clothing 15727​89
recommendation combining user social circle and fashion style 65. Waheed MF, Khalid AM (2019) Impact of emerging technolo‑
consistency. Multimedia Tools App 77:17731–17754. https://d ​ oi.​ gies for sustainable fashion, textile and design. In: Karwowski
org/​10.​1007/​S11042-​017-​5245-1/​METRI​CS W, Ahram T (eds) Intell Human Syst Integr 2019. Springer, Cham,
52. Hamad M, Thomassey S, Bruniaux P (2017) A new sizing system pp 684–689
based on 3d shape descriptor for morphology clustering. Com‑ 66. Silva ES, Bonetti F (2021) Digital humans in fashion: will consum‑
put Ind Eng 113:683–692. https://​doi.​org/​10.​1016/j.​cie.​2017.​05.​ ers interact? J Retail Consum Serv 60:102430. https://d ​ oi.o
​ rg/1​ 0.​
030 1016/j.​jretc​onser.​2020.​102430
53. Wu Q, Zhu B, Yong B, Wei Y, Jiang X, Zhou R, Zhou Q (2021) Cloth‑ 67. Jacometti V (2019) Circular economy and waste in the fashion
gan: generation of fashionable Dunhuang clothes using genera‑ industry. Laws. https://​doi.​org/​10.​3390/​laws8​040027
tive adversarial networks. Connect Sci 33(2):341–358. https://​ 68. Shirvanimoghaddam K, Motamed B, Ramakrishna S, Naebe M
doi.​org/​10.​1080/​09540​091.​2020.​18227​80 (2020) Death by waste: fashion and textile circular economy
54. Shahrabi J, Hadavandi E, Esfandarani MS (2013) Developing a case. Sci Total Environ 718:137317. https://​doi.​org/​10.​1016/j.​
hybrid intelligent model for constructing a size recommenda‑ scito​tenv.​2020.​137317
tion expert system in textile industries. Int J Cloth Sci Technol 69. Moorhouse D, Moorhouse D (2017) Sustainable design: circular
25:338–349. https://​doi.​org/​10.​1108/​IJCST-​04-​2012-​0015 economy in fashion and textiles. Design J 20:1948–1959
55. Lin Y, Ren P, Chen Z, Ren Z, Ma J, de Rijke M (2020) Explainable 70. Power DJ (2016) Data science: supporting decision-making. J
outfit recommendation with joint outfit matching and comment Decis Syst 25(4):345–356. https://​doi.​org/​10.​1080/​12460​125.​
generation. IEEE Trans Knowl Data Eng 32(8):1502–1516. https://​ 2016.​11716​10
doi.​org/​10.​1109/​TKDE.​2019.​29061​90 71. Müller O, Fay M, vom Brocke J (2018) The effect of big data and
56. Liu J, Song X, Chen Z, Ma J (2020) Mgcm: multi-modal generative analytics on firm performance: an econometric analysis consid‑
compatibility modeling for clothing matching. Neurocomputing ering industry characteristics. J Manag Inf Syst 35(2):488–509.
414:215–224. https://​doi.​org/​10.​1016/j.​neucom.​2020.​06.​033 https://​doi.​org/​10.​1080/​07421​222.​2018.​14519​55
57. Wu L, Yue X, Jin A, Yen DC (2016) Smart supply chain manage‑
ment: a review and implications for future research. Int J Logist Publisher’s Note Springer Nature remains neutral with regard to
Manag 27:395–417. https://​doi.​org/​10.​1108/​IJLM-​02-​2014-​ jurisdictional claims in published maps and institutional affiliations.
0035/​FULL/​XML
58. Becker T, Illigen C, McKelvey B, Hülsmann M, Windt K (2016)
Using an agent-based neural-network computational model
to improve product routing in a logistics facility. Int J Prod Econ
174:156–167. https://​doi.​org/​10.​1016/j.​ijpe.​2016.​01.​003

Vol.:(0123456789)