The Startup
Published in

The Startup

Fashion, Blockchain, & AI, … Oh My!

BEFORE I START: MY JOURNEY

photo by: © Shaun Fynn

From the moment we take in our first breath until our bodies are buried back into the dirt, one thing remains constant: clothes. We spend our lives here on earth with some sort of fabric on our bodies 24/7. We wake up, we put on some clothes, we go to sleep, we change into some other clothes. Rinse. Repeat. The garment industry is part of our everyday lives whether we realize it or not.

You can call it fate, or you can call it an awakening of some sort — but my life literally changed after I stumbled upon a documentary on Netflix, “True Cost”. The documentary told tale after tale about garment workers who faced modern day slavery. Once the credits rolled, I felt confused, distraught, and ultimately angry. Confused that people in North America knew nothing about it. Distraught that I never once thought about the hands that put together the garments in my closet. Angry that people could be so mistreated and no one really did anything about it. I couldn’t just wipe away my tears and move on with my life. I had to do something about it.

I opened my laptop and immediately applied for … fashion school! I decided that I was going to become an ethical fashion designer. I was going to create a new ethical clothing company.

But something didn’t feel quite right! The more I thought about it, the more I became convinced that creating a company with ethical standards would not get to the core issue of getting consumers to change their shopping behaviour. I was relying on people’s taste to create change. Instead, I wanted a concrete way to inform people to make good choices. I wanted consumers to understand that what they buy locally has a big impact on workers halfway around the world. I needed a way to connect the dots and realized: I needed to learn how to code.

This was why I began my journey back to school. I am a mature student with a Bachelor of Arts in Applied Linguistics under my belt, eagerly studying all things Computer Science.

I learned from a casual conversation about how blockchain technology is being used to trace the source of a diamond, to show whether the stone came from a conflict zone. Could I somehow implement this record-keeping / blockchain technology to make it easier for shoppers to find out whether their clothes were sourced ethically too?

I met with Professor Anil Somayaji who helped me realize that blockchain may not be the only possible solution. This is how this project came about. I devoted 12 weeks to find ways to use computer science to develop trust in supply chain management in the realm of fashion. As a Computer Security expert, Professor Anil helped me understand what was possible and pointed me in the right direction to begin developing a solution to help find a solution for the lack of transparency and sustainability in the fashion industry.

THE PROBLEM

What’s the problem with fashion, you ask? Let me break it down.

1. It literally hurts the world.

According to the Ellen MacArthur Foundation, 20% of industrial water pollution globally is from dyeing and treating textiles. The foundation also cites the International Energy Agency, stating that “textiles production accounts for significant greenhouse gas emissions… In 2015, greenhouse gas emissions from textiles production totalled 1.2 billion tonnes of CO2 equivalent, more than those of all international flights and maritime shipping combined.” Did you read those numbers?! It should scare you.

2. Some garment workers are modern slaves.

According to the Global Slavery Index, an estimated 40.3 million people were living in modern slavery in 2016, many of whom are working in the fashion industry. The index defines modern slavery as “exploitation that a person cannot refuse or leave because of threats, violence, coercion, abuse of power, or deception.” In her book, Slave to Fashion, Safia Minney explains that in the garment industry, modern slavery includes child labour, human trafficking, forced and excessive labour, and bonded labour. It includes workers who are often forced “to work through threats or intimidation … falling into debt and being forced to work for free in attempt to repay it … [and even children] who toil for 60 hours a week in garment factories and whose meagre salary is used to support their parents and siblings.” Woof.

Compounding these problems is a lack of transparency! You don’t know which companies have these problems and which don’t. Fashion is a global industry, with suppliers that supply suppliers in a long chain that makes it difficult to track who made what. It’s difficult for the company itself to track, let alone for us consumers. Woof x 2.

Currently middlemen in the fashion industry supply chain have very little accountability. Consumers can hold some retailers to account, but there is no way to check whether audits of suppliers and middlemen are credible, and it is too difficult to map all of the relationships between retailers, factories, middlemen, and raw materials suppliers.

The garment industry is well aware of these problems. In the State of Fashion 2019 report from McKinsey & Co., the need to achieve greater transparency was rated one of the top four challenges for the year ahead by fashion company executives.

The tragic collapse of an eight-story commercial building in Bangladesh in 2013 and movements like #WhoMadeMyClothes have fortunately raised more awareness for consumers like myself about unsafe labour practices. However, there has yet to be a concrete solution that is sustainable, credible, and powerful enough to incentivize big companies to be transparent.

So, what are we to do?

BLOCKCHAIN: THE ULTIMATE SOLUTION?

I began my research with the full on excitement that blockchain could be the ultimate solution to the lack of transparency in the fashion industry supply chain.

Before I dive in, what in the world is BLOCKCHAIN?! Simply put, blockchain keeps detailed records of transactions through a network of computers (no human hands involved). It stores copies of the same transaction on multiple devices, making it almost impossible for a hacker to edit or tamper with the transactional information without using enormous and improbable amounts of computing power.

The question that kept running through my mind was, could I somehow implement this blockchain technology to make it easier for shoppers to find out whether their clothes were sourced ethically too?!

I pitched this question to Safia Minney, a prominent advocate for sustainable fashion, and a human I aspire to one day meet. To my surprise (she responded to my email and yes I squealed aloud) she led me to Project Provenance Ltd., which literally “uses blockchain technology to enable secure traceability of certifications and other salient information in supply chains.” Project Provenance can help companies track everything from the raw material to the final product, so consumers can see and know where their clothes actually come from, and their environmental footprint. I then discovered several other companies (yes more than one) who are actually using blockchain to do what I was dreaming of doing!

The technology to track where a product comes from, already exists. But there’s still a challenge. There is no way of ensuring that the data stored on the blockchain is reliable. My rose-coloured view of blockchain turned when I realized that there was no real way of ensuring that the data stored on the blockchain is correct.

Luckily, I had the chance to ask Project Provenance directly about this problem in a webinar: “How can we trust the information stored on the blockchain? How do we ensure that the immutable data is correct data from the beginning?” Their response:

“It [the data] was entered by humans, so for us, it being on the blockchain doesn’t make it any more true. If it was wrong to start with, it’s just immutably wrong.”

Project Provenance’s response confirmed that there could be loads of wrong data stored on their blockchain. The fact that incorrect data can be stored on the blockchain shows that the issue is not just the lack of transparency. Instead the bigger problem is, how to filter out the wrong data and guarantee that the supply chain information is reliable.

BEYOND BLOCKCHAIN

I really (like really really really) wanted blockchain to be the ultimate solution! Alas, since it wasn’t the quick fix I dreamed it would be (sigh), I began to look into ways to filter out incorrect data prior to it being stored on the blockchain. As I write these words, I can sense the smile from Anil. He kindly (from the very beginning) nudged me to look into the world of artificial intelligence and machine vision. Of course, I was hesitant, because blockchain seemed so dreamy and fun! (Don’t get me wrong - it’s still quite amazing how data stored on blockchain is pretty much immutable) - but as Project Provenance confirmed, there could be a lot of incorrect data being stored. Tragic.

I think what Project Provenance is doing is really a huge step in creating a more sustainable and transparent world. When I asked them about how we could trust the information stored on blockchain, they told me:

therefore we’re very much about connecting networks of information together to triangulate to increase assurance that the inputs are what they say they are. That’ll be a combination of data that runs with product, but also sampling, and adding additional data points, additional trust points, proof points, like certifications. It’s about bringing all of the information together rather than one tracked thing equals true.

Project Provenance’s solution to the problem of unreliable data is to use other data in parallel to what is stored on the blockchain.

But, rather than filtering out data and using other “proof points”, I really wanted to find a pure solution that wouldn’t require sorting through immutable data and then verifying it again!

Now enter the world of automated video analytics and wireless signals! *cue The Jetson’s Theme Song * as I introduce some “futuristic” cool technology out there. (2019, you have a long way to go, but you have shown me that it’s a pretty amazing (and scary) world out there).

VIDEO ANALYTICS

Eliminating manual entry of information on the blockchain (during the different steps in the supply chain) could lead to a more transparent fashion industry. There are a few different ways to do this, including automated video analytics (enabled by artificial intelligence (AI) of course — checkout this link if you want to learn about basic AI).

The video I’d be interested in is using a closed-circuit television (CCTV) type surveillance that would be installed in the different factories of the supply chain. I’m not talking about the cameras that you see in malls and other public spaces, with a security guard (yep — a human) who analyzes the data after something terrible happens. This would be different. The solution would be installing a smart surveillance camera, aka, computerized video analytics; too many human hands involved requires a lot of monies and time, resources we don’t have much access to and, as you probably have gathered, figuring out a solution to the lack of transparency in the fashion industry is no easy feat!

Let’s define video analytics. It “involves a variety of techniques to monitor, analyze, and extract meaningful information from video streams.” The techniques that interest me most for monitoring the fashion industry are: facial recognition, movement analysis, and people counting. And the exciting part is that the technology already exists! In fact, there are tons of companies doing these types of analysis with cameras. (Some examples include: Intelli-Vision, Ntechlab, and Affectiva).

Facial recognition identifies individuals by analyzing someone’s unique facial features (of course - using AI). In China, facial recognition is being used to take attendance in schools. In a factory setting, facial recognition could serve a similar purpose. It could tell if the same garment worker is working longer hours than allowable under local labour laws. Another way that facial recognition could be used in a factory setting is identifying the age of a garment worker. Quividi, a data analytics company, uses facial recognition to detect age and gender to play targeted advertisements. If identifying the age of a garment worker is possible, it could help ensure that there are no child labourers in the factory. Can I get a “YES PLEASE?!”

However just facial recognition isn’t enough. Data retrieved from these smart cameras (analyzed by AI) could also watch for human movement. Not only did that same school in China use cameras to take attendance, they could “track what students are doing at any moment, including reading, writing or listening.” Take Axis Communications as another example, a security company that sells smart cameras to identify theft, spot intruders, or check for threats like unattended luggage in a busy airport. Honeywell Video Analytics also offers solutions for counting people and vehicles. In a factory setting, knowing how many people are on the floor could help check for compliance for fire regulations, and tracking movement could indicate whether workers are getting appropriate breaks along with the number of hours they’re working in a day.

Potential Problems

All of the potential solutions get me excited, but nothing is perfect. There could be unintended consequences that could be brought about through retrieving so much data from the factory. Smart surveillance scares people, and rightly so.

Privacy is the biggest concern. There are tons of people who get creeped out (and they should) because video analytics is pretty smart and can get pretty scary. Take Amazon Go for example, their smart shopping experience uses similar cameras to analyze your every move. Even without facial recognition, Amazon can predict behaviours (with AI) and encourage you to spend beyond your means through targeted advertising. You may feel like you’re in control, but using all the data they collect, they basically help you change your mind. SCARY! In a factory setting, using video analytics could put workers in a constant state of stress as they avoid behaviours that could be interpreted negatively by managers. On top of that is the constant threat of hackers and state surveillance. If you want a thought provoking read, see this article about the potential pitfalls of AI enabled surveillance.

You know the saying, “garbage in, garbage out”? Well, AI is only as good as the data you feed it. A terribly sad example of this is with Amazon (again). Recent research by the MIT Media Lab revealed that Amazon’s facial recognition technology (Rekognition) had a more difficult time identifying gender in darker-skinned and female faces! According to an article on The Verge, “Rekognition made no mistakes when identifying the gender of lighter-skinned men, but it mistook women for men 19 percent of the time and mistook darker-skinned women for men 31 percent of the time.” If the algorithm is biased against darker-skinned women, it may misidentify the length of time that individuals have been working in the factory, leading to false readings and potential bias against certain workers.

The knock against AI is always that it is resource intensive, and video analytics gives the AI a lot of data to analyze. It takes a lot of computing power to process, making it potentially really expensive. Since fashion is such a cost competitive industry, the resource demands of video analytics may make it a little less appealing.

WIRELESS SIGNALS

Another possible solution is to use wireless signals! This would reduce privacy concerns, potentially at a lower cost.

According to a paper from MIT, wireless signals can identify moving objects by, 1) comparing reflections from the receiver with static objects, or 2) using the human body as an antenna and tracking the resulting beam. This technology has been around for quite a while, but all of the data was too complicated to interpret before we had the processing power to use neural networks. In a Vice article, the author of the same paper later explained that they were able to train “a neural network to interpret the way radio wireless signals bounce off a person’s body and translate it into the movement of 14 different key points on the body, including the head, elbows, and knees.” In fact, Cognitive Systems Corp uses this same technology for home security. Instead of using cameras in a garment factory, a wireless system could do pretty much the same thing that the smart surveillance could do. It can count the number of employees and determine whether they’re taking breaks using an AI algorithm that is fine tuned to monitor the garment industry and connect it directly to the blockchain. Cool!

Another super cool thing about wireless signals is that you can detect emotion! The same group from MIT developed a new technology that “can infer a person’s emotions using wireless signals. It transmits an RF signal and analyzes its reflections off a person’s body [including breathing and heartbeat signals] to recognize his [or her] emotional state (happy, sad, etc.).” Their research shows an 87% rate of accuracy in predicting emotions. That’s pretty amazing. This means that if this technology could be used in a factory, it could detect the emotional state of the garment workers. You could even create a rating system to alert auditors to elevated levels of stress in the workers.

THE FIGHT FOR THIS TECHNOLOGY

So, why aren’t we doing anything?!

The fashion industry has a lot of problems. Problems so big they seem almost impossible to tackle. Blockchain is pretty great at record-keeping, but it can’t be used alone. Since we’re talking about transparency and trust, we simply cannot trust the human hand that enters the data. That’s where superhero AI comes in. Video analytics can be a great way to ensure that garment workers are working in good conditions, only working appropriate amount of hours, and making sure that there are no child labourers (ever!). Sadly, no technology is ever perfectly dreamy, and video analysis has the potential to be costly and unattractive to people who find cameras creepy. Wireless signals could help with the cost and privacy issues by detecting motion and doing almost the same thing that video analytics can do! Wireless signals are pretty new place for applying AI, but they seem like a promising way forward toward a more transparent supply chain in the garment industry.

Let me say this loud and clear, THIS IS NOT A NEW TECHNOLOGY! There are tons of companies that offer solutions in this space. Sadly, none are specifically applying it in the realm of fashion. Why is there no-one out there, doing anything?! I honestly don’t know.

It’s not because the extra costs aren’t justified. Harvard Business Review explains that “[c]onsumers are voting with their dollars — against unsustainable brands.” Products with a sustainability claim accounted for 14.3% of the market in 2013, but 16.6% in 2018. More and more often, if consumers are given the choice between a sustainable and non-sustainable brand, they are going to choose the more ethical choice. There is a whole lot of money in the world of fashion. According to Fashion United, a business platform for the fashion industry, the “global apparel market is valued at 3 trillion dollars, 3,000 billion, and accounts for 2 percent of the world’s Gross Domestic Product (GDP).” The money is there!

There is technology out there that can create an actual concrete solution that is sustainable, credible, and powerful enough! Does this get your heart pumping as fast as it does with mine? Amazing! Reach out and contact me, and let’s make some things happen.

Let’s be the change, use these ideas, and develop a solution to make the fashion industry transparent, sustainable, and ethical, so that workers that are part of this giant industry are treated fairly.

--

--

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store