217 – Charlotte Fanneau of Heuritech – Finding Trends in Images

Detecting what’s inside millions of untagged images shared on social media for fashion & luxury houses to measure product trends & know clients better... Charlotte Fanneau, Product Marketing Manager for Heuritech (AI that empowers fashion e-retailers to bridge the gap between social and commerce with cutting-edge deep learning technology), joins Pavan Bahl, Marc Raco, and Ilan Tito at Shoptalk Europe in Copenhagen. Powered by Sennheiser. In this episode: The journey of getting in the door as a new quantity in geographical areas until Heuritech is more known Instead of a data API, how Heuritech worked in building the best product plug and play for teams, fitting the daily needs of brands How Heuritech was created by two PhDs in machine learning, they dove into fashion industry because of so many images Knowing what mattered in those pictures, training the model to detect many type of details, and recognize products from specific brands, especially luxury bags Then training the machine to recognize that element in any new image, then scale it in social media Heuritech’s strength in detecting specific SKU for brands From hours to a few days to pull many SKUs and training machines against images already on Internet, or even coordinate photo shoots for new products launching Tracking metrics, the number of times seeing specific products in images How Heuritech’s tech is good for metrics companies and niche influencers Tracking all luxury, analyzing that and selling to specific clients who are not necessarily luxury, identifying what trends are, and selling that data back Forward thinkers with luxury brands, helping the rest of the world to take more risks with data to back intuitions Getting past hesitations based on past experiences brands have had Because visual search has not been very good until now, how it is easy to show brands that they can succeed, which solves issue of brands not identifying products in social media images, as they can now be identified, and being able to funnel into intelligent content The possibility of identifying products and brands in video, but the data challenges before employing a strategy The many possibilities, collaboration with LOUIS VUITTON Source material to identify the items and referencable with various body types, ways clothing is draping, and the need to build a reference system The goal to be key partner with fashion and luxury specialists Revealing and helping people grow what they’re good at, contrasting fashion between Copenhagen and Paris, and electric drums and singing