Continuing on our promise during Inman Connect NYC 2017, we are now moving our image advisor service out of the labs into general availability. It has taken us over 18 TB of images and 3.2 nerd-years to train our neural network to perfection.
Sally – our Convolutional Neural Network (CNN) based deep learning platform for computer vision, has been trained specifically on residential real estate imagery and is now available as a service for anyone to leverage. Sally can look at a set of images for a home and automatically select and recommend the best picture to be your thumbnail and also reorder your photos for the best viewing experience for consumers.
Sally uses a growing library of millions of images available via PropMix to constantly train itself. Sally can now analyze each image and identify not just the type of room – kitchen, living, dining, etc. – but can also recognize what is available in the image. For instance, Sally can easily recognize the image of a kitchen along with the type of flooring, countertop, cooking amenities, shelves, sinks etc. Sally provides a score-based ranking of its recognized objects enabling the higher level algorithms to make contextual decisions for property listings, valuations, and/or fraud detection.
Expect a lot more to come from Sally in the coming months to empower sellers, buyers, agents, and appraisers in the real estate market. Please subscribe here for exciting news on Sally.