AI and Tree Coverage

A frequent question is “How does Deep Lawn’s AI handle tree coverage?” and we are going to answer it here.

One of the great things about Deep Lawn’s AI is how it handles tree coverage and its multiple tools to attack the different challenges presented by tree coverage. The most common ones are

• Individual Trees

• Tree Lines

• Heavy Canopy

Individual Trees

Individual trees are handled with the pattern recognition baked into our AI model. Deep Lawn has measured over 1 million properties in the US and Canada and of course - many of those properties contain individual trees. Our AI model is trained to assume that individual trees will most likely have turf underneath them because that is usually true.

AI Measurement of an individual tree in the front yard

In this example - the AI saw a tree in the front yard and assumed there was turf underneath it. And, if we check how the AI did using the Google street view of that property - we can see that is correct.

Google street view of the same property

Tree Lines

Tree lines are common - particularly in backyards. And unlike individual trees - tree lines do often indicate the treatable turf area has ended. Deep Lawn’s AI takes this into account. When the AI notices multiple overlapping trees it will - for lack of a better term - make a judgment call - that the turf extends a short distance into the tree line and then stops.

Now we all know that AI can’t make judgment calls and what is actually happening is an algorithm estimating the tree line based on the previously measured properties. But saying a judgment call sounds a lot better and gets the idea across.

AI measurement of a property with a Tree Line in the backyard

The AI determined that the tree coverage in the backyard probably blocked off a portion of the turf and made a cut-off. Also, these tree line judgments can always be adjusted either by you internally - or by your customers on the storefront to make sure they are precisely dialed in. 

Heavy Canopy

Heavy canopy is the hardest to deal with and for these Deep Lawn’s AI needs to use HD imagery. HD imagery is available through NearMap and uses aerial imagery rather than satellite. It is also updated every 3-4 months. While HD through NearMap does cost more - it is just the thing for dealing with a heavy canopy.

Deep Lawn’s AI will measure off of both the most recent image of the property AND the most recent “leaves off” winter imagery to see through the trees. 

Standard Imagery (left) and HD Measurement (right) of a property with a Heavy Canopy

As you can see the leaves off HD NearMap imagery allowed the AI to see the turf and features in the backyard. Much of this could be missed with standard imagery.

Conclusion

Deep Lawn’s AI has many tools for dealing with tree coverage and you - or your customers - can always edit those measurements to ensure they capture exactly what you are looking for.