Check out these presentations from DIL postdocs for more inspiration at 4pm after class!

Reading / Tech

Please review our very own project pages!

Tentative plan

First, Catherine Burton from Urthecast will give us a brief overview of earth observation satellites, remote sensing image processing, machine learning algorithms, APIs and the advent of big data analytics and unstructured data.

Then we’ll have class review!

Then, we’ll have a lively review of the semester. If a team can’t make it for the final presentations, this is an alternative date. Teams on deck so far:

The carpool project

Data project evolved; challenges:

Currently, the data collection has to be done as a 2-step process:

Overall goal is to reduce competitive ride-sharing apps prices by reducing ride costs to a fraction of the competitive prices, even if that means walking 20 minutes for the rider.

Earth observation and big data analytics (Catherine Burton)

Background in geospatial analytics “Urthecast” : They have 4 satellites, including a color live camera

Remote sensing: the process of obtaining information about an object and gathering data from a sensor from a distance

The technology started during the war, cameras were strapped to pigeons or balloons

1946: first image taken from a rocket 1960: first meteorological satellite 1960-72: spy satellite program 1972: first civilian land satellite 1999: first high resolution image satellite

Different technologies

NASA’s EO (earth observation) system

Aerial photography:

Kite photography:


SAR (synthetic aperture radar)

Measuring the Earth with sensors

how to visually interpret satellite imagery?

The four V’s of big data:

Volume Velocity Variety Veracity

Geographic Information Systems (GIS)

Big data and IoT have been merging together

Pattern recognition: algorithms apply some filters to an image and apply morphological recognition Machine learning: algorithms aimed at prediction of data beyond recognition only Deep learning: algorithms extract high level complex abstractions as data representations through a hierarchical learning process