A Japanese retailer was looking for ways to improve its product matching to help customers find clothing products from their own personal photos. Remi AI used image recognition, feature extraction comparisons and other measures to identify similar clothes in their digital store and other online stores from clothes photographed on social media and on locally taken photos.
The method developed was able to group over 1,000,000 different clothing products available on a range of major e-commerce stores into a significant number of clusters.
The final solution allowed users to submit photos taken in local conditions and the model would find the most similar clothing option available online.
Rapid image recognition pipeline for Product Matching.