The Slyce company has revealed several enhancements over the course of 2015. One of these was launched during the 2015 cycle and is known as the enhanced 3D object recognition system. This solution allows the system to recognize any product in any category that exists in the retailer’s catalogue. This phenomenal solution was only launched by Slyce for those retailers who operated in verticals. Categories like home decor, tools, appliances, electronics and fashion are considered verticals. This allowed the user to take his or her own images, known in layman terms as User Generated Content images or for short UGC.
The way the recognition works is in two unique ways. The first approach is via a fully automatic visual search as well as a human assisted flow. In simple terms the image is first processed by the fully automated recognition flow and if this does not provide substantial results in attempts the human assisted flow. This flow allows the user to provide the algorithm with hints, this all in real time, this will allow the algorithm to provide the user with the most relevant set of results specific to the reference image that was provided by the user.
The first flow known as the fully automated flow uses a statistical matching approach where the image is compared to an existing extensive database of already indexed images. This also includes several images of the same product but taken from a variety of different angles so that no image is left to chance. The human assisted flow on the other hand rely’s on existing search mechanisms provided by the retailer. This extracts potentially relevant products in real time and then allows for a polishing process to extract the items that match up to the image provided. Using this approach of two flow processes allows the Slyce platform to match products to any UGC image from any category and any retail catalogue.