Classify Homes

using supervised machine learning

Throw prop data at a machine and it learns

Works well!

Not so much if the image is 

interior,

view obstructed, or

not even a home, but that is true for a human too...

We can classify images as we get them

And store the prediction

This data in 

Current Search

waaaaaiiiiiting

Both in Results Searching for "Victorian"

Should the search be all NEW? And visual?

like Instagram

Why this matters to the next phase

Do we add more classes (gather more data) or do we fine tune (adjust the learning) the 10 we have?

More classes means more data gathering

Fine tuning means more learning

Thanks

Both options are good!

But in the interest of time, this is the perfect time to choose direction