CoreLogic NZ has a long and proud history in geospatial excellence. It’s not just about our tech either. We’re proud to have some people in our team who can actually remember hand-drawing maps for the NZ Government. To this deep human expertise, CoreLogic is adding the latest in deep learning technology and applying both to our extensive imagery, geospatial and property datasets to gain new insights and learn more about our land and property.
One such example is how we’ve taught machines to automatically recognise swimming pools through imagery analytics.
CoreLogic actively supports local government organisations across NZ with data requirements, and swimming pool compliance is one of those areas. For councils to be able to check swimming pool compliance, they have to know about them in the first place. And, whether due to the cost implications of compliance or just plain ignorance of council by-laws, not everyone proactively tells them. Surprise, surprise.
This is where geospatial analytics can really shine. For example, we had previously mined a dataset of rating and building consent information to identify around 40,000 swimming pools on New Zealand residential properties. However, we knew there must be more, and thanks to the CoreLogic imagery service (which provides cloud-free coverage for 100% of NZ mainland addressable properties), we could see swimming pools that weren’t in the ‘official’ records. How could we scan the whole country to identify these ‘missing’ pools even though they’re easy to identify from imagery? The answer was to turn to some of the latest techniques in image analysis.
In short, our humans taught our machines to identify the pools. Or, to get all technical about it: using the CoreLogic imagery service, we taught our deep learning neural networks to identify swimming pools and estimate pool sizes and locations. Using our geospatial smarts, we then linked these pools to parcels and properties. Additional algorithms were used to identify temporary pools and attribute our estimates with confidence scores.
The result was impressive. The original dataset had originally identified 40,000 pools, but an additional 18,000 pools were found with a ‘high confidence’ rating, far closer to the estimated total number of pools in the country. This provides CoreLogic clients with the most complete and current data available on swimming pool locations, providing intelligence for local and central government, emergency services, insurance providers identifying both increased risk and improving sum insured estimates, and an untapped market potential for pool product suppliers. As they say, knowledge is everything.
To see how our geospatial and imagery analytic capabilities can help your business understand its own market better, or if you want to know more about our unique swimming pools dataset (which for each residential property includes a Yes/No ‘swimming pool present’ status, X & Y (or Latitude/Longitude) coordinates of the swimming pool location, indication on swimming pool area, a confidence score, a ‘temporary pool’ indicator, an image tile and image metadata), contact us on 0800 355 355.