• Real estate marketplace leader

    Building a product recommendation engine using Computer Vision

    How a leading architectural agency used visual search to understand 1.5M user generated images.

  • Background

    A real estate marketplace leader offers an online platform that connects architects working on commercial-scale buildings with the manufacturers they need. Their platform hosts over 1.8 million images that users rely on to make purchasing decisions and get inspiration for their architecture projects. Every month users add over 30,000 new images of recently completed residential, commercial, and institutional buildings.

  • Information

    Use Case

    Visual Search, Product Recommendation

    Industry

    Real estate

    Client

    Real estate marketplace

Details

Challenge

With user-generated content playing a significant role in helping buyers connect with sellers on their platform, the company needed a solution that would provide a highly scalable system for exploring content and images to increase user engagement and boost conversions in their online marketplace. Three years of testing every image recognition technology led the company to choose Clarifai

architizer-home-on-water

 

Solution

In the company’s ecosystem, buyers are typically architecture firms and sellers are building materials manufacturers. Buyers or architects on the platform rely on user-generated content and images of architecture materials for purchasing decisions and design inspiration. Because architecture is such a visual and design-driven medium, surfacing content based on images is the best way to serve relevant recommendations to users.

To recommend relevant image-based content to its buyers, the company used Clarifai’s custom training and visual search solutions. They were able to detect patterns and visual similarities in a wide array of architecture-related photos and make content more discoverable on their platform. Using Clarifai’s custom trained models, the company was able to create a custom image recognition model to understand architecture-related features like “facade system,” “cantilever,” “brutalist,” and “living wall.” These custom concepts allowed them to personalize their image recognition solution and categorize their images in a way that was most relevant to their business, rather than relying on a one-size-fits-all general computer vision model.

Results

With Clarifai's visual search capabilities, the company was able to automatically understand 1.5 million user-uploaded images on their platform and use that information to surface visually similar content to users. This content helped connect architects with materials they felt inspired by or might want to buy.

30%

Reduction in website bounce rate

2.25 minutes

Average time spent on website pages

1.5 million

Images reviewed for quality checks and duplicate removal

  • The custom training was the deal clincher, but overall, Clarifai’s level of communication and insight has been hugely helpful to our initiative.”

    Chief Product Officer

  • Ready to get started?

    Whether you're a start-up or a Fortune 500, contact us to discuss how to enrich your data with AI assisted labeling to annotate datasets accurately, quickly and affordably.

    Learn more about Clarifai

    Schedule a demo

    Discuss your solutions options

  • Already a customer?

    If you are encountering a technical or payment issue, the customer support team will be happy to assist you.


    Contact support >