Fashion AI in action, season II

A while ago, we talked about the emerging trends of putting AI into fashion (see link here) and identified few superstars in this field. It seems these superstars are swelling and Fashion AI moves faster than we have expected. Today we dish out the latest progress again.


First, AI catalog automation 

Last time when we first identified this Indian background startup, it just focused on online visual search and personalized recommendation. But now its website has fully grown under the central theme of “AI automation and personalization”, offering a very wide service package.

And above all, the most exciting breakthrough: world’s first AI-based human model generator, an end-to-end merchandising and catalog automation system. We know fashion retailers usually have to conduct a continuous stream of photo shoots with real human models for their product lines for every new season or new run. Although such process has been in certain degree simplified and digitalized in the past, it is still not cost-effective. It is said solution can analyze a clothing image laid out on a flat surface, then automatically generates a human model of any size, shape, skin color, or with any kind of shoes, while also predicting how the garment would fit. If other solutions in the industry use 3D body scanners or more, approach needs no special hardware and uses in-house algorithms to help the AI generate human body types and visualize the garment on it. The days of a headless model photographed against a white background might be over.

Check out below demo:


Invested by the most elite VC firm Sequoia, is already selling the auto-tagging solution to Zilingo, a Sequoia funded e-commerce marketplace with its presence in over 10 countries. It is very likely will keep leveraging the partnership with Zilingo and testing the human model generator technology on its platform.


Second, Findmine: AI complete the look

Shoppers might want to buy a complete outfit or always try to figure out the mix&match, but in reality, it is not that feasible to manually create a complete outfit for every single product for the whole fashion inventory. Findmine solves this bottleneck by applying AI algorithm to determine the best outfit suggestion for a customer that is on-brand, in-stock, and seasonally appropriate. It is said that FindMine can put together 11 million outfits on daily basis. On top of that, the client testimony claims the solution is a plug&play implementation. See video demo:


Third, Heuritech: AI Fashion trends forecast 

We have to say this French startup was born with a silver spoon in its mouth; meaning they signed two of the biggest luxury fashion houses on the planet as its first clients. Not many startups can have such privilege and this could act as a door opener to entice other fashion brands. Heuritech’s core strength lies in predicting fashion trends based on the analysis of millions of images shared every day on social media. By doing so, it truly enhances brand’s product design and merchandising planning in advance. We revisited this startup again because it is leaping forward very quickly and pushing the expansionary plan into the US and Asia. We shall keep keen eyes on its development, so stay tuned.

Author: Cecilia Wu