Clothing styles can be detected via video feed image analysis

27-04-2018 |   |  By Rob Coppinger

The clothes people wear will one day identify them to retailers as potential loyal shoppers with the help of video camera feeds, machine vision and biometric identification.

The detection of clothing styles, the colour and type of jacket or blouse or trousers, for example, is a way of determining someone’s age and gender, which are key data points for retailers’ marketing departments. A teenager is unlikely to wear a tweed jacket, for example. Brick and mortar retailers want to know who it is who walks past and only looks in the window or decides to enter and buy something, or not. Internet retailers have the advantage of cookies and information given by a visitor’s registration to build up that all important picture.

“In the future we’ll be doing that [clothes classification]. [Detecting] that jacket that means you’re a male of a certain age,” said Aura Vision Labs co-founder, Daniel Martinho-Corbishley. For now, Martinho-Corbishley’s start-up, Aura Vision Labs, is using full body analysis for determining age and gender using intelligent machine vision. “We’re doing proof of concept projects with a number of…brand retailers, one of them has about 150 stores in the UK and some are on Oxford Street.”


The video feeds can come from a shop’s existing security cameras. Gait analysis, examining how people walk, enables the computer to identify gender and age when an individual has their back to the camera. A picture of the face is preferable, but not necessary as partial facial images are enough. The image analysis, Martinho-Corbishley explains, works in almost any lighting condition. The analysis of an individual’s journey into a shop and through it, where they dwell with products that they may or may not buy is referred to as marketing, advertising and technology, or MadTech.

Martinho-Corbishley and his colleague Jaime Lomeli are the founders of Aura Vision Labs. They claim their firm’s MadTech can capture up to 100 per cent of in-store visitor demographics, dwell spots, walk-ins and walk-bys. This data could be accessed and analysed by retailers through a cloud-based platform. Martinho-Corbishley points out that shops can already detect the mobile phones of people who enter the shop, but this provides very limited information, for example they will try to estimate an individual’s gender.


Martinho-Corbishley and Lomeli developed the technology while doing doctoral electronics and computer science research at the University of Southampton. Martinho-Corbishley submitted his final thesis in early April. The two students were studying in Southampton’s electronics department’s vision, learning and control research group.

Martinho-Corbishley and Lomeli have raised £100,000 seed investment through London-based MadTech investment firm, Collider. Aura Vision Labs was launched last year when the two students pitched their idea at an on-campus Dragons’ Den style investment competition. They received early-stage support from the Web Science Institute’s Z21 Innovation Fund and Martinho-Corbishley and Lomeli attended the Consumer Electronics Show (CES) in Las Vegas with the help of the University’s Future Worlds incubator. At CES, they met retailers who have since become partners for the proof of concept testing.

By Rob Coppinger

Rob Coppinger is a freelance science and engineering journalist. Originally a car industry production engineer, he jumped into journalism and has written about all sorts of technologies from fusion power to quantum computing and military drones. He lives in France.

Related articles