Industry Technologies
By Yvonne Heinen-Foudeh – Chief International Correspondent
What had been named Artificial Intelligence represents a module in a whole bundle of game-changing technology tools at our fingertips that provide chances for our human species to get up to speed. While Part 1 of our series (see Needle’s Eye # 8 - February 2022) provided insight into how operational practice modules and applications summarized under the abbreviation AI will help to get closer to the illusion of zero defects and to reach a greater degree of increased precision.
Countering the Overproduction Menace
Furthermore, we rolled out how clever and mature algorithms fed with correct data, already today support fashion brands to get ready for personalized products and intelligent forecasting.
Merchandising and manufacturing (in that order) apparel, home textiles, furniture, and other textile goods more will experience impact over the coming years and certainly beyond: Technologies leveraging AI or bundling with it will enable decision-makers to create and verify the right solutions for essential problems in the first run instead of wasting energy and resources through the analog trial methodology. Instead of losing precious time. Time we may not have anymore.
For the textile and apparel industry as well as for retail pre-dominantly predictive applications are making a name for themselves at this stage. Although broadly offered as a service for instance by Google Cloud this data analysis use is still far from being state-of-the-art in the apparel arena.
What an AI application does in that context is to provide information on individual and target group-related customer preferences that are determined based on clicks, shopping baskets, orders, internal, external search queries, and return behavior. These preferences enable the adaptive recommendation algorithm to develop personalization rules in real-time. The door to on-demand fashion is thus wide open.
To ensure that the customer does not end up in a filter bubble and the offer becomes too narrow, smart recommendation technology uses explore mechanisms. The system specifically deviates from the learned patterns to recommend products that sometimes surprise the customer.
One of the biggest challenges remains the complex issue of sizing. European studies indicate that fashion, sportswear — and all kind of garments and shoes that do not satisfactory fit — causes between 35%–55% of returns with online shopping and that is for all age groups.
Similar figures were collected for reasoning by consumers – not only female ones – leaving stationary retail without purchasing a piece of garment or an outfit.
The good news
A new generation of computer scientists has taken up the subject to the next level. Some start-ups or university spin-offs are achieving promising approximations concerning the right size and, through questioning, also covering individual fit expectations. (Among them the start-up Presize featured in the NEEDLE’S EYE Issue #7- January 2021, see intro and interview with CEO and co-founder Leon Szeli.)
And How About Production?
It is the combination of AI with robotics (to “AI robotics”), which upgrades the plain robot, previously known as a more mechanical C-driven tool to one being able to learn, to plan, to reason, to represent knowledge, develop solutions, and even create visions. And with this type of intelligent manufacturing system the robotization of the highly complex joining process for flexible materials — ergo also the sewing of two-dimensional cut parts into three-dimensional shells — is (literally) within reach. And not only that, industry-ready systems are already available on the market.
Softwear Automation Inc., as an example, today after nine years of R&D offers the first fully autonomous Sewbot work lines based on the use of AI. “Our Sewbots take cut fabric pieces as input, then puts them through a series of automatic steps, and output a finished t-shirt,” Palaniswamy ‘Raj’ Rajan, Chairman & CEO of the Atlanta-based advanced start-up explains. That application to combine AI, proprietary advanced robotics, computer vision, and IoT [internet of things] technologies is enabling on-demand production at scale and a reduction of transport costs, waste of precious sources as well as emissions.
To what extent this technology will enable the replacement of human sewers and provide the opportunity of nearshoring on a broad scale is yet to be determined by the critical mass required for investment into re-shored production of apparel.
With one of Softwear Automation Inc.'s Sewbot work lines, a t-shirt can be made and dispatched within 48 hours.
AI meets AR
Fashion in the Metaverse to be the next hot topic combining AI with AR (Augmented Reality) and not at the horizon but just around the corner: Sports and footwear apparel brand New Balance recently announced to join Nike and Adidas in the metaverse with three pending trademarks. Italian premium brand Prada already is present in Mark Zuckerberg’s brave new (fashion) world. Last month at NRF (National Retail Federation) conference in New York, Ralph Lauren-CEO Louvet Patrice confirmed to be chasing opportunities in the metaverse.
AR + AI: with the Fit Analytics acquisition Snapchat is getting ready to turn the smart phone into a virtual changing room. Photo: dpa
From past to present into future
Perception and intelligence have fascinated researchers for thousands of years. After the Second World War digital computers emerged, a new phase began. Scientists all over the world began to focus intensively on computers and intelligence. AI research was born.
What became famous was the conference at Dartmouth College in Hanover, New Hampshire, at which leading scientists discussed 1956, where leading scientists discussed how to use programs to develop Artificial Intelligence programs. At this conference, many aspects of artificial intelligence were defined that are still being discussed today. Starting from the Dartmouth conference, a kind of working definition came into use: “AI is the science that deals with the development of machines that behave intelligently”. The M&A activities of some relevant tech players are pointing the way forward for the digital fashion journey with the possibilities of artificial intelligence. Originally dedicated to helping people share images of their lives with their social circles on its flagship app Snapchat, Californian Snap Inc. acquired the previously mentioned Fit Analytics (for US$124 million) in March 2021. The young tech company as well based in Berlin, Germany’s Mecca for the start-up scene, has built technology to support shoppers finding the right-sized apparel and footwear from online retailers. Along with a wider set of personalization tools and other analytics, the aim is to help retailers figure out how to sell more overall. Thus, Snap is adding not just a technology team — there is a staff of 100 at the company site in the German capital, now reporting to Snap VP of Engineering Nima Khajehnouri — but a substantial e-commerce technology business into its portfolio. 3D garment simulation specialist CLO Virtual Fashion Inc., headquartered in Seoul, Korea invested in PixelPool from Den Haag. That start-up company stands behind Dtail – a 3Dcloud-based visualization platform. “PixelPool provides one of the best 3D fashion experiences through its various products such as virtual showrooms,” said CLO-CEO Simon Kim. The CLO investment will go towards the continued development of Dtail, already enabling fashion brands to plan, visualize, and sell their collections by linking product data to 3D virtual samples and synchronizing those assets to a realistic, customizable virtual environment which elevates the visual merchandising and hroB2B sales processes.
What became famous was the conference at Dartmouth College in Hanover, New Hampshire, at which leading scientists discussed 1956, where leading scientists discussed how to use programs to develop Artificial Intelligence programs. At this conference, many aspects of artificial intelligence were defined that are still being discussed today. Starting from the Dartmouth conference, a kind of working definition came into use: “AI is the science that deals with the development of machines that behave intelligently”. The M&A activities of some relevant tech players are pointing the way forward for the digital fashion journey with the possibilities of artificial intelligence. Originally dedicated to helping people share images of their lives with their social circles on its flagship app Snapchat, Californian Snap Inc. acquired the previously mentioned Fit Analytics (for US$124 million) in March 2021. The young tech company as well based in Berlin, Germany’s Mecca for the start-up scene, has built technology to support shoppers finding the right-sized apparel and footwear from online retailers. Along with a wider set of personalization tools and other analytics, the aim is to help retailers figure out how to sell more overall. Thus, Snap is adding not just a technology team — there is a staff of 100 at the company site in the German capital, now reporting to Snap VP of Engineering Nima Khajehnouri — but a substantial e-commerce technology business into its portfolio. 3D garment simulation specialist CLO Virtual Fashion Inc., headquartered in Seoul, Korea invested in PixelPool from Den Haag. That start-up company stands behind Dtail – a 3Dcloud-based visualization platform. “PixelPool provides one of the best 3D fashion experiences through its various products such as virtual showrooms,” said CLO-CEO Simon Kim. The CLO investment will go towards the continued development of Dtail, already enabling fashion brands to plan, visualize, and sell their collections by linking product data to 3D virtual samples and synchronizing those assets to a realistic, customizable virtual environment which elevates the visual merchandising and hroB2B sales processes.
AI development – long-term future
Artificial intelligence (AI) includes different tasks such as image recognition, voice recognition, and natural language understanding, to name but a few. "While these tasks were challenging a few years ago, they are now considered easy tasks thanks to the recent advances in machine and deep learning," explained Dr. Mohammed Abufouda.
The Data Science expert, himself involved in R&D for various industrial applications, further stated: "Having said that, all of these tasks fall under so-called Narrow AI – AI systems that can perform one specific task under predefined context and scope." In comparison to Narrow AI, then General AI would enable human-like cognitive and intellectual tasks without human intervention. Although this type of AI is still not existent at the moment, an optimistic estimation for developing it will not be before five decades from now", Dr. Abufouda explained to The NEEDLE’S EYE.
.jpg)
Asking the experts: Dr. Mohammed Abufouda consults company for and develops AI applications in demanding industrial environments.
A bit of etymology
Where might the term algorithm have originated?
Donald E. Knuth can be considered the most famous algorithmist. In his book "The Art of Programming" he reports that early linguists derived the term from the Greek álgiros (painful) and arithmós (number). A derivation that probably says more about the linguists and their attitude towards the fearsome algorithms.
The word was probably so difficult to remember because it derives from the name of the Arab mathematician 'Al-Chwarizmi'. He wrote a mathematical textbook in the 9th century. When this was translated into Latin three centuries later, the name of the author was adapted to the supposed root from Latin or ancient Greek.
Source: Katharina Zweig, Algorithmen haben kein Taktgefühl (Algorithms have no tact)
Game changer AI: Much of what is in the pipeline seems astonishing, some are frightening, most of it encouraging. It is in our hands what we decide to make of it. Photo: Pavel Danilyuk
Virtual shopping at 3D boutique created w. PixelPools platform – Photo: Dtail.
ARTificial Fascination: Media recommendations
Let’s face it, artificial intelligence opens the possibility of any digital emulation and with that imitation - of anything and everything: of human voices, music, any creation – including fashion styles and shows – of images of any kind, be they still or moving images – video materials. And yes, with it the gate is wide open to every form of manipulation. Already where we stand today, in case of any doubt, audio-visual materials are to be taken with a grain of salt. ‘Deep fake’ to the term for this AI application.
There are dangers of misuse but also gigantic opportunities for preservation and cultural advancement. For example, the digital reconstruction of the ancient oasis city of Palmyra in the governorate of Homs with the first finds from the Neolithic period, destroyed during the Syrian Civil War is progressing.
Another example of unlimited possibilities: D-ID from Tel-Aviv, Israel, with their ‘MyHeritage’ app turns classic family photos into lifelike moving portraits and enables transforming still photographs into ultra-realistic video, capable of saying whatever one wants to say.
Here are some sources, recent events, and enlightening lectures for those users within the NEEDLE’S EYE community who want to dive deeper, and are trying to imagine and understand what’s possible – our future with AI.
- Book recommendations:
- Big Data – Die Revolution, die unser Leben verändern wird, Kenneth Cutier, Victor Mayer- Schönberger
- The Art Of Computer Programming (TAOCP), Vol. 1: Fundamental Algorithms Donald E. Knuth, 3rd Edition; Publisher, Addison-Wesley, US; language: English
- Algorithms Have No Sense Of Tact – [Algorithmen haben kein Taktgefühl] – Katharina Zweig, Publisher: Heyne-Verlag; language: – to date German only
- Experience AI: Futurium Museum, Berlin/Germany:
- Exhibition “Arts and Its Doubles”, EPFL Pavilions and University of Lausanne, Lausanne/Switzerland – until May 1, 2022. https://epfl-pavilions.ch/exhibitions/deep-fakes-art-and-its-double
- And in conclusion, one more AI tidbit: Get smashed and inspired of what will be – what already is possible by watching exemplary animation of celebrated Mexican paintress Frida Kahlo, re-enacted with the help of an actress explaining one of her most famous works, “The Wounded Deer” from 1947. https://www.d-id.com/museums-and-ai-letting-art-tell-its-story/
What if artists, scientists and thinkers, long-past could explain their work themselves? AI software app from Israel makes them come to life again, like e.g. Mexican artist, Frida Kahlo. ©: D-ID