ERICSON HERNANDEZ: And I'm Ericson.

ALEJANDRA VASQUEZ: And we're using

machine learning to find potholes

on the streets of Los Angeles.

ERICSON HERNANDEZ: LA has over 7,400 miles

of roads and freeways.

ALEJANDRA VASQUEZ: That's a lot of roads

and a lot of potholes to fix.

ERICSON HERNANDEZ: Right now, construction workers

have to manually inspect roads for dangerous potholes

and cracks or rely on people calling in tips,

which takes a lot of time.

ALEJANDRA VASQUEZ: While studying at LMU,

we started to see if we could figure out

a faster way to identify potholes throughout the city.

ERICSON HERNANDEZ: First, we needed data.

So we rigged a camera to a car, and we drove around

to capture footage of different roads and freeways.

ALEJANDRA VASQUEZ: Then we used TensorFlow,

Google's open-source machine learning tool,

to develop a model that could quickly identify potholes, road

cracks, and other weird stuff with a high rate of accuracy,

which means workers can spend less time finding potholes

and more time fixing them.