Electricity Access in Developing Countries from Aerial Imagery

Project Summary

Boning Li (Masters Electrical and Computer Engineering), Ben Brigman (Electrical and Computer Engineering), Gouttham Chandrasekar (Electrical and Computer Engineering), Shamikh Hossain (Computer Science, Economics), and Trishul Nagenalli (Electrical and Computer Engineering, Computer Science) spent ten weeks creating datasets of electricity access indicators that can be used to train a classifier to detect electrified villages. This coming academic year, a Bass Connections Team will use these datasets to automatically find power plants and map electricity infrastructure.

Themes and Categories
Year
2017
Contact
Ashlee Valente
Center for Applied Genomics and Precision Medicine
ashlee.valente@duke.edu

Project Results: The team gathered electrification ground-truth data for over 36,000 villages in the Indian state of Bihar, and also collected measurements relevant to electricity consumption for those villages including lights at night data and irrigation metrics. They also created an Amazon MTurk tool that crowdsourced the annotation of key electricity indicators (such as power plants and transmission lines) in imagery data.

Partially sponsored by Bass Connections and the Duke University Energy Initiative

Click here for the Executive Summary

Faculty Leads:

Kyle Bradbury

Leslie Collins

Timothy Johnson

Marc Jeuland

Guillermo Sapiro

Project Manager: Boning Li

"The Data+ team created two new datasets that we'll immediately deploy as a part of our core research efforts and will serve as the basis for an upcoming Bass Connections in Energy project. The outputs will be used towards two new research projects on energy infrastructure and access in developing countries, and will serve as the ground truth data for developing machine learning techniques for identifying energy infrastructure and access. The students were fantastic - hardworking, passionate about their work, and all-around wonderful people to work with." — Kyle Bradbury, Lecturing Fellow and Managing Director, Duke Energy Data Analytics Lab

 

Related People

Related Projects

Producing oil and gas in the North Sea, off the coast of the United Kingdom, requires a lease to extract resources from beneath the ocean floor and companies bid for those rights. This team will work with ExxonMobil to understand why these leases are acquired and who benefits. This requires historical data on bid history to investigate what leads to an increase in the number of (a) leases acquired and (b) companies participating in auctions. The goal of this team is to create a well-structured dataset based on company bid history from the U.K. Oil and Gas Authority; data which will come from many different file structures and formats (tabular, pdf, etc.). The team will curate these data to create a single, tabular database of U.K. bid history and work programs.

Producing oil and gas in the Gulf of Mexico requires rights to extract these resources from beneath the ocean floor and companies bid into the market for those rights. The tops bids are sometimes significantly larger than the next highest bids, but it’s not always clear why this differential exists and some companies seemingly overbid by large margins. This team will work with ExxonMobil to curate and analyze historical bid data from the Bureau of Ocean Energy Management that contains information on company bid history, infrastructure, wells, and seismic survey data as well as data from the companies themselves and geopolitical events. The stretch goal of the team will be to see if they can uncover the rationale behind historic bidding patterns. What do the highest bidders know that other bidders to not (if anything)? What characteristics might incentivize overbidding to minimize the risk of losing the right to produce (i.e. ambiguity aversion)?

In this project, we are interested in creating a cohesive data pipeline for generating, modeling and visualizing basketball data. In particular, we are interested in understanding how to extract data from freely available video, how to model such data to capture player efficiency, strength and leadership, and how to visualize such data outcomes. We will have four separate teams as part of this project working on interrelated but separate goals:

Team 1: Video data extraction

This team will explore different video data extraction techniques with the goal of identifying player locations, ball location and events at any given time during a basketball game. The software developed as part of this project will be able to generate a usable dataset of time-stamped basketball plays that can be used to model the game of basketball.

Teams 2 & 3: Modeling basketball data: offense and defense

The two teams will explore different models for the game of basketball. The first team will concentrate on modeling offensive plays and try to answer questions such as: How does the ball advance? What leads to successful plays? The second team will concentrate on defensive plays: What is an optimal strategy for minimizing opponent scoring opportunities? How should we evaluate defensive plays?

Team 4: Visualizing basketball data

This team will work on dynamic and static visualization of elements of a basketball game. The goal of the visualization is to capture information about how players and the ball move around the court. They will develop tools to represent average trajectories be in these settings that can also capture uncertainty about this information.

Faculty Leads: Alexander Volfovsky, James Moody, Katherine Heller

Project Managers: Fan Bu, Greg Spell, 2 more TBD