Improve quality of information we posses about our mother earth.
What it does
It analyses images from satellites and drones and extracts business related data (counts sheeps or cars for instances).
How we built it
As a sample application we choose a task of counting amount of cars available for sale in cities of Berlin and Munich. We used satellite images provided by airbus to extract this information using image recognition algorithm based on IBM Watson neural network for image recognition. Here maps were used both to prefilter the data and thus optimise the speed of processing and to show the resulting information to the user.
Challenges we ran into
There are a lot of technological issues starting with the current state of input data to the algorithms available for data processing and analysis. The satellite's pictures have too low resolution for Machine learning, so we had to recognize the blocks of cars.
Accomplishments that we're proud of
Calculated the number of cars parked by the car dealer in Munich
What we learned
We had experience with integrating different API's from HERE, airbus, watson, training neural network.
This prototype can be extended to approximate the coal, iron, other resources stored on factories. This information is important for stock market. With increasing quality of pictures it can be used to measure speed of transport and measure anomalies of average speed, which can be a sign of decreasing road's quality.