Can we get hyper-local infrastructure time series with StreetView and deep learning?
This article is originally published at https://robertgrantstats.wordpress.com
More Google StreetView ideas. Suppose you wanted a measure of infrastructure investment, or of fragility because maintenance has been cut back? How about some of those wacky instrumental variables that economists love? But you want it at hyper-local level. Well, maybe you can track this stuff visually. I picture some convolutional neural network that takes images like these, identifies common objects — road signs in this case, which are uniform and regulated so should be easy to pick out — and measures the extent of some problem — here, being overgrown with vegetation. I saw this very sign a few weeks back when approaching the Segensworth roundabout, a landmark of Southern England on the same scale as Stonehenge or Tower Bridge. This is no backwater but a major road and a major intersection. The direction to Fareham is completely hidden now (fortunately, nobody goes to Fareham; they only leave). I thought to myself that the obscured sign — not the only one by a long way — was a marker of Austerity Britain, and then the idea came to mind.
We see here peak credit crunch in Nov 2008, then April 2011, September 2012, April 2015, May 2017, May 2018. I saw it in May 2019 and it looked pretty bad.
You could presumably detect graffiti, potholes and broken windows and such as well. Given the lack of a public API, you’d have to break the Google T&Cs and have some poor devils doing remote piecework for you screen-capturing these images (obviously you should not do that). I leave programming of this as an exercise to the reader, and if you land a great job as a result of playing with it, remember to buy me a beer one day.
Images (c) GoOgle. Conflict of interest statement: the author is “from” Winchester and may not have been entirely objective in his assessment of Fareham.
Thanks for visiting r-craft.org
This article is originally published at https://robertgrantstats.wordpress.com
Please visit source website for post related comments.