Fypio is one of our customers that develops an app that helps you search for your perfect home in a revolutionary way. They make use of hundreds of data sources, connecting them to the location of your future homes to help guide you in finding what homes best fit you.

They wanted a web based engine to visually tag real estate images based on entities in the image. They needed a system with high performance and usability as hundreds of thousands of images are tagged manually by an offshore workforce, as well as a stable and fault tolerant one to be able to handle poor communication lines in Bangladesh. Main features:

  • Visually identify entities in an image and set tags accordingly
  • Be able to get an overview to of all tagged images for confirming and correcting tags
  • Be able to use advanced filters to view all images for specific tags with boolean operators
  • Be able to store color information for each image, and to find images with similar colors
  • Be able to store and search for millions of images on the backend with good performance
  • Advanced statistics to follow up quality and speed of tagging

screencapture-ec2-54-84-169-106-compute-1-amazonaws-com-3000-carousel copy

screencapture-ec2-54-84-169-106-compute-1-amazonaws-com-3000-batchstats

Technology used:

  • Node.js
  • Mongo DB / mongoose
  • Client side javascript, html and css
  • Ajax for server communication
  • Google Charts API

System architecture, system development in a scrum environment, communication with off shore team in Bangladesh:

Joakim Ahlén