Problems and prospects of digitised technologies in fashion

Smart clothing and smart garments have a significant contribution to health and well-being and quality of life through the integration of IoT along with AI.

The fashion industry is a contributor to the United Nations sustainable development goals (SDGs) for social, economic, and environmental sustainability, and it needs to achieve the following major SDGs goal: build resilient infrastructure, promote inclusive and sustainable industrialisation, and foster innovation; Goal :ensure sustainable consumption and production patterns; and goal: take urgent action to combat climate change and its impacts.

5. AI in fashion
   In this section, we discuss the significance of AI in the fashion industry for multiple applications such as prediction of health-related issues of elders, patients, and children, fashion trend forecasting, and dress recommendation based on environmental parameters.
   5.1. Trend forecasting and dress recommendation
   Clothing is a type of sign that conveys the wearer’s interior perception through the exterior appearance. It can transmit information about the wearer’s preferences, faith, personality, occupation, social status, and outlook on life. Predicting fashion and recommending the clothes is possible by analysing the fashion information available in social networking through images and text. Furthermore, there are several fashion-related forums, such as Chictopia and Loo-book, which are recognised for sharing personal fashion styles. The characteristics and qualities of these clothes photographs shared on social media can reveal more about the wearers’ personalities. Although researchers have investigated social media textual content such as post and comment prediction, emotion, and information diffusion, research using social media picture analysis is still limited in evaluating fashion styles or trends. Online clothes images from social media and other sources, on the other hand, can be an excellent source for assessing and building online fashion recommendations.
   Consumer fashion choices are frequently influenced by a variety of elements, including demographic, regional, individual preference, interpersonal influence, age, gender, season, and culture. Effective recommendation systems are critical tools for successfully doing e-commerce operations in the current day. In the case of fashion recommendation, the use of convolutional neural networks (CNN) or deep learning (DL) methods for image analysis, statistical analysis for recommendation system comparison, quantitative or mixed-method based on research design, and the formulation of experimental models for machine learning (ML) application can be an essential integration for creating an efficient fashion recommendation system. Reference concentrated on the study of patterns for various consumer groups using finely grained fashion elements. For the initial period, a large-scale fashion trend (FIT) dataset built from Instagram reports and usage statistics was offered, and a Knowledge Enhanced Recurrent Network (KERN) model was developed, which employed deep recurrent neural network capacity to model time-series details. Reference proposed a data-driven computational abstraction method based on an AI algorithm to improve the reliability of image-based data processing and reduce the cost of fashion image processing. Reference used DL algorithms to investigate fashion styles and trends across different populations based on street fashion photographs gathered from various Internet sources. CNN is a collection of deep neural network machine learning methods that are commonly used by researchers to extract features from large amounts of visual picture data. The statistical data of fashion product visuals, social media participants, and randomised targeted people from image-sharing web pages have been evaluated using different CNN models to analyse and visualise fashion trends based on street style photos. Information was collected from Instagram postal data by using the CNN model to identify the most prominent top fashion accessories in a certain location. This outcome provided a framework for businesses’ decision-making awareness, such as examining customer patterns in various locations, product penetration, and finding common products. Reference used the AlexNet CNN architecture for image classification and feature extraction to create a deep user-based efficient recommendation system by comparing the homogeneity of the user vector with the image vector. Deep net theory has been used to determine clothing style and proposed three improvements to deep architectural systems in the distribution of the computational world driven by deep learning’s robust classification capacity and ability to process a large data volume in the big data age.
   5.2. Health prediction with smart clothing
   In the smart cloth, the number of health sensors that are embedded in the cloth assists in obtaining real-time health data of that particular person. The smart cloth can be utilised for real monitoring of health and other activities of the baby, patients, and older persons. In the case of a baby, the sleeping pattern, health conditions, and milk feeding pattern monitoring are crucial to maintaining the health of the baby. In addition, the real-time tracking of babies and their activities is significant to ensure safety and security. In the case of the patients, the biological sensors integrated into the smart cloth of patients keep tracking of pulse rate, body temperature, oxygen intake, and function of neural signals in the real-time scenario. In the case of an older person, health monitoring and real-time tracking of their activities are vital to ensure safety and health. Early fall detection, fall detection alerts, medicine intake remainder, and emergency alerts are the major requirement of smart cloth for older people.
   As discussed early, AI has gained wide attention in the prediction of events based on input data available in the form of images and texts. In smart cloth, the real-time sensor data (heart rate, heart rate variability, temperature, oxygen intake, and stress level) obtained from the sensory unit is the input to the AI model for the prediction. DNN has received a lot of interest in AI because it can filter input through a cascade of numerous layers, with each consecutive layer using the output from the preceding one to inform its results. DNN necessitates minimal data pre-processing. Many of the filtering and normalisation activities that must be performed by human programmers when utilising other machine learning approaches are handled by the network itself.
   A framework of smart cloth with DNN and cloud server for detection of health abnormalities and generating emergency alerts has been formed. Here, the smart cloth is embedded with a wireless communication protocol to transmit the sensor data based on time interval on the cloud server gateway. The DNN model is applied to the sensor data that is available in the cloud server. Based on the outcome of the DNN model, the necessary further action is represented. In case of health abnormalities, it sends an alert to the personal doctor/health provider, and in case of emergency, it sends alerts to the family members to take immediate action. In addition to the smart cloth, footwear like shoes is also integrated with sensor and communication-based systems to monitor the walking pattern and other health parameters.
6. Blockchain in fashion
   Block chain was developed for business and supply chain applications as a private block chain, as it provides privacy and controlled access to approved and identifiable participants. Depending upon the accessibility level, each participant has the right to access the subset of the information. Private block chain encourages building trust and transparency among the participants in the supply chain. Furthermore, each player can sustain their strategic advantage without exposing all facts and methods to competing organisations. Conversely, the block chain shall capture all transactional data and provide customised access to supply chain participants while remaining auditable and verifiable. The participants that are involved in the fashion supply chain, include fibre producers, yarn manufacturing, fabric manufacturing, apparel manufacturing, retailer, and customer.
   The private block chain-based framework of the fashion supply chain is proposed in, where it comprises organisation-level and operation levels. The actions engaged at the organisational level include methods for personalised accessibility of records, the configuration of private block chain infrastructure for the supply chain, and techniques for connecting the block chain network. The process of recording and storing information connected with various supply chain stages will be detailed at the operational level. For supply chain track and trace application areas, transaction validation mechanisms and smart contracts can also be set up. Maker aspects of the organisation-level include channels, applications, ordering services, and membership services.
   Channels are network subdivisions used for information segmentation, and it has their block chain, which is recorded on a distributed shared ledger. A group of upstream suppliers, often trading with a single product, might link up to a single channel or a group of channels, every for a separate kind of operation. Peers are process methods that handle the digital contract, and depending on their functions, they enable the creation, endorsing, and validation of transactions. Only the primary actors of a supply chain must present peer systems based on their roles because it requires capital and system management.
   Applications are user-responsive software interfaces that can be leveraged to execute block chain queries, and to include in the shared ledger, queries can be made to read, update, and authenticate. The block chain application can be executed on numerous devices with unique access rules from each partner. The ordering facility is often a third-party service that is in charge of network maintenance. Its major purpose is to accumulate a collection of operations complemented by supply chain associates, enroll them on a block, organise them in a rational sequence, and deliver the block for authentication. In a private block chain network for supply chains, membership services produce and distribute new public key pairs to participants. The key combinations define the kind and scope of exposure to the block chain network based on the function and ownership of supply chain partners.
   At the operational level, the design of a traceability architecture is crucial. The architecture should exhibit operations such as gathering, planning, organising, and trading traceability data at various stages of the fashion supply chain. Most upstream participants in the fashion supply chain accept raw materials in various types as an entry from providers and execute diverse strategies to create the end product that is forwarded on to the succeeding supply chain associate. This procedure is repeated by multiple supply chain associates until the end product is delivered to the store. It is also critical that the relevant data is gathered and disseminated to comply with local rules and certification bodies’ standards. To simplify complexity, the traceability information was divided into four categories: public shareable, private shareable, linkage, and secured information. Public sharable information can be distributed to any third party and customer since it contains critical information that must be captured. The amount to which public shareable information is collected and shared determines the level of transparency of a supply chain for its client.
   Private shareable information is a collection of information that is critical for B2B operations, particularly those that are valuable for material processes. When this information is securely exchanged among supply chain partners, it can improve visibility and minimise risk. Secured information is encoded data that can only be accessed by a chosen group of B2B partners. Secured information is sensitive and confidential, and it includes financial data, intellectual property data, and a range of other pieces of information that can be utilised to obtain a significant gain. Linking information is critical to the block chain’s operation, and it is this data that is checked throughout each transaction across the shared ledger managed by the smart contract. This includes four value fields: Transaction Signature, Public Key, Traceability ID, and Asset Value. To validate the transaction, all of these would be checked against the global state of the shared ledger.
7. AR and VR in fashion
   In this section, we discuss the significance of AR and VR in the fashion industry for providing a real-time experience. Here, initially, we discuss AR, and it is followed by VR.
   7.1. AR
   In fashion commerce and industry, AR comprises virtual try-on that use personalised or non customised simulated models to replicate the presence of apparel product patterns on a human form. A recent study proposes a better innovative virtual trial room integrating IoT mirrors to provide customers with a more accurate try-before-you-buy clothes purchasing practice, where AR was employed for skin color identification and outfit suggestion based on skin color. Customers in the fashion industry frequently seek to locate adequate clothing that meets their requirements, and in a few instances, the patterns, colors, and designs elements can make them feel uncomfortable. To overcome these challenges, numerous researchers advised following a user-customisation system in which the client can design his or her garment. For that, has suggested using AR to explore textiles with realistic illumination. In their study, they used an AR application to create real-time augmentation on garments by identifying spare cloth in video footage. Research has confirmed that real-time 2D AR on a nongrid object like clothing can monitor a modern circular system and apply the proper radiance and shades to the product.
   In mass customisation, they created a whole new product mode by combining customised and mass production to provide clients with a variety of brands, draperies, materials, and colours. Scientists are seeking to manufacture footwear with motion detectors rather than modifying t-shirts. Reference established a systematic approach for designing and customising footwear for youngsters, which supports shoe personalisation and pattern development. The design module’s functions include color, embroidery, texture, and carving design, with unity 3D utilised for colour and texture representation and the UVW mapping technique used for mapping. Reference suggested a full-featured garment customisation system based on AR, where OpenGL 3D rendering, Azure Kinect somatosensory technology, and a somatosensory virtual fitting room are employed.
   The user can apply numerous designs to the cloth and then try virtual fitting, where it encapsulates the deepness data stream of human bones via the Azure Kinect somatosensory technology and does bone-tracking handling as well as 3D clothing virtual try-on in the right location on the human body. Research has been conducted on marker-based AR utilised in a physical clothing marketing ecosystem via AR mobile applications, which cooperate with the consumer by generating data on the product, such as colors available, size, and stock, and visualising a 2D pattern of the item. A mechanism is offered to use the 3D model of a t-shirt by scanning a catalog as an enhanced version of the above approach. Both AR applications were created using the Unity3D and Vuforia AR kits.
   7.2. VR
   Online shopping has improved dramatically with the growth of the Internet over the past few decades. As a consequence of the pandemic, buying goods and services via Internet platforms has grown commonplace. As a result, it is projected that a new form of shopping environment based on VR would become increasingly prevalent. A VR environment would promote multisensory and physical engagement of fashion-related retail operations, shopping service content to consumers, presenting more precise and realistic products, and facilitating a genuine buying capability. There has yet to be a systematic investigation on the nature of experience supplied to customers while shopping for fashion products using VR. User experience (UX) is a highly complex topic that combines human emotions, usage conditions, and expectations. This physical paradigm allows for the exploration of product events in practically any dimension, including aesthetic, cognitive, and emotional components, as well as sensory aspects. Utility, fun, aesthetics, usability, intention to use, and impairment concerns are among the UX criteria they proposed. Unlike VR games or sports, VR shopping is not a platform that can be enjoyed by a large number of people at the same time. The attributes of UX experience in terms of aesthetics, intention to use, playfulness, usability, and utility has been illustrated.
   VR is a computer environment that permits customers to immerse themselves in a virtual world via multiple forms of virtual sensory synthesis response, such as hearing and sight. It refers to cyberspace in which users can experience a sensation of existence and absorption, as if they were physically present in that area, in an intentionally created world of senses. Immersive VR is a method in which a person employs and observes a head-mounted display (HMD) attached to a computer or mobile device. The user’s field of view is regulated, the visual stimuli of the actual world are prevented, and the user only receives a 3D image generated by the HMD by putting a distinct device in an immersive VR environment. In this way, it becomes possible to experience a virtual world as if it is the real world, and an intense sense of reality and immersion occurs. However, immersive VR has a few limitations like headache, dizziness, and nausea, and also user’s need to buy VR devices negatively impacts customers for product shopping using VR. An empirical experiment was undertaken focusing on the UX rating items created in this manner. UX of immersive and non immersive VR is evaluated under the parameters such as telepresence, sharpness intention to use, and playfulness. The immersive type has high UX than the non immersive type focused when averages of telepresence, sharpness intention to use, and playfulness are evaluated. It concludes that the fashion product shopping experience in the immersive VR empowers users to experience telepresence, sharpness intention to use, and playfulness parameters effectively.
8. Discussion and recommendations
   According to the UN, the fashion industry needs to meet the following goals: build resilient infrastructure, promote inclusive and sustainable industrialisation, foster innovation, ensure sustainable consumption and production patterns, and take urgent action to combat climate change and its impacts. To meet all these goals, the fashion industry needs to incorporate digitalised technologies like IoT, AI, blockchain, AR, and VR. With this motivation, this study explored the different studies that implemented these technologies for smart cloth (health), supply chain, circular economy, dress recommendation system, fashion trend forecasting, health prediction, and virtual and augmented based shopping experience. From our exploration, the following findings and suggestions are addressed:
   (i)Sustainable Material Usage. It is concluded by the United Nations that the fashion industry is also one major sector that contributes to carbon emissions due to its manufacturing practices. The manufacturing unit of the fashion industry needs to adopt 3D printing in designing clothes and garments with sustainable materials. 3D printing has gained attention in the footwear industry to mass customisation insole, outsole, and middle sole of shoes. It is encouraged to use 3D printing in the other areas of the fashion industry for the benefit of achieving a sustainable manufacturing environment as it enhances time efficiency, zero-waste production, product customisation, and design flexibility.
   (ii)Wide Adoption of Blockchain in Fashion. The fashion industry comprises the supply chain that presents the stage-by-stage production of products from raw material to the final product. The main concern in the fashion supply chain is the protection and privacy of a huge volume of product data that are available with industries. Currently, many industries have effectively implemented the blockchain for their supply chain. So blockchain technology needs to be widely implemented in the fashion industry for better security and transparency. Along with this, blockchain technology achieves an effective circular economy that enhances trading platforms.
   (iii)Advancement in Energy Storage. In smart clothing, energy storage and energy management play a significant role in the functioning of IoT devices embedded into it. The researchers of material science need to do thorough research on the energy storage devices that can be embedded inside the cloth with flexibility and comfortability. Along with energy storage devices, the communication protocol integrated into the smart cloth should consume low power and have minimum RF effect on the human body.
   (iv)Advancement in Technology Training. AR and VR have gained wide attention in the fashion industry, as a retailer and manufacturers have implemented these two technologies in terms of virtual fitting, in-store navigation, virtual try-on, and so on. The technology training of the employers through simulation should be effective and have minimum consequences so that they can learn the utilisation of these two technologies in real time for different operations.
   (v)Integration of IoT, AI, and Edge Computing. At present, in this study, we have mainly focused on smart clothing in the health sector; as in the current scenario, health is to be given utmost importance in terms of patients, babies, and older-aged individuals. From the exploration of the previous studies, it is concluded that limited research is carried out on the integration of edge computing, AI, and IoT in smart clothing for the real-time identification of anomalies in health.
   (vi)Smart Clothing-Based Framework for Rescue Operation. In the rescue operation, the location tracking and identification of the precise person to get assistance plays a crucial role. A framework needs to be designed in such a way that the people can easily identify the respective rescue staff with location on the interactive LED display. Here, people and rescuing staff should wear smart clothing, as the information related to rescued individuals and their location will be wirelessly displayed on the LED display. This indeed enhances the tracking of individuals during the rescue operation, and also this concept can be replicated in other areas for different interactive applications.
9. Conclusion
   The fashion industry generates 3 trillion dollars and contributes to 2 per cent of world GDP. In addition to this United Nations concluded that this industry has a lack of concerns for social and environmental issues in terms of generating CO2 emissions and wastage in material consumption and production. However, the integration of digitalised technology like IoT, AI, block chain, AR, and VR can achieve the United Nations SDGs relevant to the fashion industry. With the motivation of this, this study explored the progress of digitalised technology in the fashion industry. During the exploration, the study focused on digitalising technologies in the fashion industry for smart clothing, forecasting fashion trend, dress recommendation on the basis of environmental conditions, prediction of health, real-time supply chain, and fashion and shopping experience. Based on the exploration, this study presented the limitations and suggested recommendations such as wide adoption of blockchain in fashion supply chain; advancement in energy storage for smart cloth; integration of IoT, AI, and edge computing; and smart clothing-based framework for rescue operation for the future enhancement.

(This is the second part of the article; the first part was published in the November edition.)

About the authors:

Dr N Gokarneshan is from the Department of Textile Chemistry, SSM College of Engineering, Komarapalayam, Tamil Nadu, India.
Sona M Anton is from the Department of Fashion Design, Hindustan Institute of Technology and Science, Chennai.
B Padma, R Hari Priya, AJ Abisha Raju, S Kavipriya, M Karthiga, are from the Department of Costume design, DR.SNS Rajalskshmi College of arts and science, Coimbatore.
S Umamageshwari is from the Department of Fashion Design, CSH, SRM Institute of Science and Technology, Kattankalathur, Chennai.
Usha Kumari Ratna is from the Department of Textiles and Clothing, Avinashilingam Institute of higher education for women, Coimbatore.
V Sathya is from the Department of Fashion Design, SRM Institute of Science and Technology, Ramapuram, Chennai.
P G Anandhakrishnan is from the Department of Fashion Design, Saveetha College of Architecture and Design, Chennai.