Stuart Nathan talks to entrepreneur Bella Trang Ngo, who is applying AI and machine vision technology to the problem of ill-fitting bras
We often say at The Engineer that engineering intersects with people’s lives in many ways that are overlooked by the public, particularly when reporting on efforts to inspire young people to consider engineering as a career. But Bella Trang Ngo is taking that to a level we have not encountered before. While it’s true that clothing design and manufacture is an overlooked sector of engineering, Ngo is using some of the most advanced weapons in the engineering armoury to ensure that one particular item of clothing fits better. Namely, bras.
Considering the gender make up of UK engineers (just 12.37 per cent are female) it’s likely that many people reading this will not realise how serious or widespread the problem of ill-fitting bras is.
According to Ngo, around 80 per cent of women wear bras that do not fit properly, and the consequences can be severe. It’s not just a matter of the discomfort we all encounter when clothes don’t fit properly. If bras provide inadequate support, whether this is because they are too large or too small, the mechanical stress and loading this causes on the body can lead to poor posture, chronic back and shoulder pain and even migraines.
“The loading goes onto the shoulder muscles and that directly affects the neck, and all the nerves that pass through it into the central nervous system,” Ngo explained.
“We are talking about 70 per cent of women who experience breast pain on almost a daily basis that is not cancer related, for example. And that leads to self-confidence and self-image issues on top of the physical problems. Just all because of a poor fitting bra.”
Ngo’s response to this has been to work with fellow students at UCL to develop a system based on machine vision and artificial intelligence to ensure that women can find a bra that fits correctly. A business studies student herself, Ngo has worked with Prashant Aparajeya, a machine vision scientist at Goldsmiths University, to design the system which is now at the centre of a company she has founded called Brarista. The endeavour recently earned her a highly commended award in the Royal Academy of Engineering’s Launchpad competition, entitling her to mentoring and training support through membership of the Academy’s Enterprise Hub.
The business model behind Brarista is that manufacturers and retailers pay to use the service, which prospective customers then access through a smartphone app. They are then directed to put on what they believe to be their comfiest, best-fitting bra and take photos of themselves in front of a mirror (or get somebody else to take them) in several specified positions. They also fill out a short questionnaire on the price range in which they wish to shop, and the sort of outfit they are expecting to wear with the bra they are looking for.
These images are then fed into the system’s algorithms, which analyse them in the same way that a trained bra-fitter would, to asses whether the bra the customer is wearing actually fits them properly. It does this by encoding the skills of a bra fitter who works by sight, rather than by using a tape measure. “We can’t disclose how many photos a user needs to take or how the algorithm works, because we are still in the process of obtaining a patent for it,” Ngo said. “But we have a specialty and expertise and understanding of how the breast shape and size and the bra size correlate and we have worked on building a data set that is proprietary to us to train our system to make sure that what we are producing is at grade.”
Ngo herself is a professional bra-fitter. “When I was doing a Masters in technology entrepreneurship at UCL, the nature of the course is that you have to find a real world problem that can be solved by technology, I was looking at different things that can be done to help a woman with bigger breasts, because I’d heard that 80 per cent statistic, which I was a bit cynical about; I thought it was just to get women to buy more bras.
“I did some market research and completely randomly, I met a professional bra-fitter who worked at Rigby & Peller at the time, it’s the shop that used to fit the Queen’s family. All she had of me was a picture of myself on Facebook and she asked me what size I was wearing. When I told her, she said not only was I in the wrong size but I was five or six sizes smaller than I should be wearing. I was in disbelief, thinking there’s no way that’s the case. We met up and she brought me the size I never thought I would ever be and I never felt better. It literally made me feel like a different woman. The way I stand, the pain of my neck and my chest area all improved almost instantly.”
It was this encounter which led to her being trained as a bra-fitter and also inspired the project which led to Brarista. The most common method of bra fitting in High Street stores, using tape measures, is only accurate in about 30 per cent of cases, Ngo added.
Ngo is conscious of the privacy concerns that might worry potential customers and assured The Engineer that none of the photographs used by the system are stored and are deleted after processing. The customer receives a list of bras from Brarista retailer partners that, according to the algorithm, would be a good fit, and the option to buy one online
Ngo is Vietnamese and UCL’s innovation enterprise development sponsors her to stay in the UK on an exceptional talent tier 1 visa. “They also supported me with free coworking space in Kings Cross and mentorship support as well,” she said.
Two weeks after winning the Enterprise Hub membership, Ngo had not yet been matched up with a mentor but said that she had had contact with the Gammon family who sponsor the RAEng award, who had indicated they would be keen to work with her on an ongoing basis to support the project. The Enterprise Hub mentorship is scheduled to begin in January.
“The motivation that keeps me in the business is I truly believe that women deserve a better way of shopping for bras offline and online,” said Ngo. “It’s time that we make sure that everything that women buy works for their body. More than 80 per cent of women’s clothes bought online are returned, so that is a lot of money and a lot of time lost on the end consumer side, but it also has an environmental impact. Systems like this – and there are an increasing number of them – could help significantly reduce that impact and wastage of time and money.”