As an extension of its ‘Shop the Look’ Pins, which identify specific items within any Pin image and connect users to purchase pages for each relevant product, Pinterest is now working on a new option called ‘Complete the Look’, which will take into account the products you’ve searched for and provide related recommendations, based on relative trends and other factors.
As explained by Pinterest:
“Complete the Look leverages rich scene context to recommend visually compatible results in Fashion and Home Decor Pins. Complete the Look takes context like an outfit, body type, season, indoors vs. outdoors, various pieces of furniture, and the overall aesthetics of a room, to power taste-based recommendations across visual search technology.”
As you can see from the above examples, the process essentially broadens its recommendations to visually similar or contextually related products, based on the aforementioned factors.
Pinterest trained its model based on a dataset of images which included complete scenes of varying product matches. Pinterest then cropped out the specific products, which enabled the system to recommend similar matches based on what it had learned were relevant partners. As with all algorithm systems, it’s fairly technical, but you can read more about the process specifics here.
Based on this initial training model, Pinterest’s systems can now identify relevant product matches based on image context – here are some examples of its contextual recommendations.
The process will help connect more Pinterest users to relevant product matches and recommendations, which could help to further fuel its growing eCommerce ambitions. And with Pinterest leading the social media pack in terms of product discovery, the addition could prove significantly beneficial for users as they seek out just the right companion pieces in various contexts.
Complete the Look is currently being tested internally, and will eventually be made available within Pinterest’s recommendation tools.
This latest Social Media News has been posted from here: Source Link