In the AI Startup of the Week, the editorial staff of ai.nl is featuring promising AI-startups, their innovations, solutions and challenges. In this first episode: Amsterdam-based Pixyle.ai who is making automatic tagging and visual search easy for third-party fashion retailers.
Artificial Intelligence, as a transformative technology, will probably take a few more years before reaching its true potential. Before it gets there, however, AI is transforming a number of industries in ways that were not previously understood. From construction, packing to retail and healthcare, AI is leaving its mark.
One of the industries where it has been showing its transformative power is fashion and online commerce. In the world of fashion, online stores might seem easier to manage but they tend to struggle more than physical ones when it comes to managing their product catalogue. Amsterdam-based Pixyle.ai wants to help these online fashion brands with their product taxonomy using AI-powered automatic tagging feature. Here is a look at the idea and how it works.
AI-powered automatic tagging for fashion industry
In the world of online commerce, a well defined product taxonomy could be the difference between success and failure. A well structured product data means that fashion retailers can improve their efficiency, enhance the customer experience and even grow their business. Most online retailers tend to struggle with their product taxonomy and Pixyle.ai provides an API that allows any brand to integrate visual search into their digital properties.
This is possible thanks to image recognition software capable of detecting fashion items in image. The startup with €100,000 in funding is using its deep learning algorithms to instantly identify objects and transform them into reliable and accurate product data.
The biggest advantage of AI is its ability to work with a large amount of structured or unstructured data. In the world of product taxonomy, which used to involve people manually inputting data, AI can make the process up to 90 per cent efficient.
AI can be used to build a well defined data structure of fashion categories supporting various methods such as categorising, organising, and even classifying items systematically. This is where Pixyle.ai comes in and uses an AI-based approach to enable automatic tagging that eliminates manual product tagging.
Here is how it works: the AI algorithm works by examining an image and identifies features associated with particular keywords, which are part of a predefined fashion taxonomy. The use of deep learning means that the algorithms can process the labelling faster and with automatic tagging, the products can be organised and images tagged faster.
The repeated use of the algorithm leads to the AI algorithm building a metadata for catalogue assets and eventually tag items within fractions of seconds.
“AI technology is only as good as the data it is trained on. So that’s why I put a lot of focus on creating a massive database and knowledge base for fashion,” says Pixyle.ai founder and CEO Dr. Svetlana Kordumova, in an interview with VentureBeat.
Visual AI to solve a complex problem
Pixyle.ai’s visual AI is capable of processing and tagging an entire fashion catalogue in no-time. It does this with the help of computer vision and deep learning image tagging models. This model automatically generates highly accurate product attribute information necessary for retailers to catalogue their product online.
The visual AI approach does not stop at generating relevant tags for different designs but also supports effective organisation. The deep learning model produces rich and detailed taxonomies. When combined with a retailer’s product, the result is improved filtering options, faster product discovery, textual search options and even recommendations.
The underlying technology may not seem novel to anyone who uses a smartphone to point at images and get recommendations immediately. Pinterest has been built on this same computer vision-power search technology while Google has brought its computer vision tech to Android devices with Google Lens.
Pixyle.ai sets itself apart with its efficiency, speed and accuracy. The startup is also targeting third-party fashion-focussed retailers who are in need of a visual search system for clothes, accessories and even jewellery. With over 200,000 images processed and turned into 300 categories, Pixyle.ai has trained its algorithm on colour and patterns.
The visual AI search from this Dutch startup ensures retailers don’t have to build their own dedicated image search system. With technology changing fast and fast fashion eating the retail world, Pixyle.ai’s visual search shows the prowess of AI in the fashion industry.
To find out more about Pixyle.ai, go to https://www.pixyle.ai/