Economics From Afar

Artificial Intelligence, Economics, Satellite Imagery

WHO WE ARE

We at ITU are studying satellite and aerial imagery to develop tools that will assist Government and Non-Government organization in analyzing urban population, road structure, urban and rural structural development, agricultural regions, animal migrations and destruction caused by natural disasters. Primarily we have studied satellite and aerial images in order to detect residential areas using computer vision and machine learning techniques.

NEWS

  • ITU faculty members win Google Research Scholar Award – THE NEWS
  • Google research award for two ITU teachers – DAWN
  • Project is In-progress

What's The Idea

To overcome the challenges of lack of information and regular surveys, machine learning based satellite imagery algorithms have been proposed to predict the economic indicators of a region. These indicators (ranging from poverty estimation, slum detection) not only help government design policies but are vital tools for the businesses to understand their customers, design their business strategy and evaluate their business model. Unfortunately, these algorithms’ predictions are not interpretable and being based on homogeneity, they fail to generalize over the regions. Instead of popular black-box techniques, we plan to create an interpretable economic well-being analysis using satellite imagery and geo-spatial datasets (e.g. estimating the density of buildings, closeness to parks, population prediction, etc.) and solve the domain adaptation problem by constraining over the interpretation rather than just generic image or feature level adaptation.

Faculty Members

Dr. Mohsen Ali Principal Investigator
Assistant Professor, Department of Computer Science

mohsen.ali@itu.edu.pk

https://itu.edu.pk/faculty-itu/mohsen-ali/

Dr. Izza Aftab CO-PI
Assistant Professor and Chairperson, Department of Economics

izza.aftab@itu.edu.pk

https://itu.edu.pk/faculty-itu/izza-aftab/

COLLABORATIONS

FUNDINGS

Our Team

Aliza Masood
Aliza Masood Research Associate
Zuhha Azhar
Zuhha Azhar Graduate Fellow
Sohail Danish
Sohail Danish Graduate Fellow
Object Detection, Domain Adaptation
Muhammad Fasi ur Rehman
Muhammad Fasi ur Rehman Research Associate
Sad Akbar
Sadaf Akbar Research associate
NOOR UL ISLAM
Noor ul Islam Research Associate

Projects

USING SATELLITE IMAGERY FOR GOOD

Government of Punjab has been revamping vaccination program to increase its geographical coverage and to better manage vaccination staff spread across whole Punjab. To analyze the coverage of the vaccination activities especially in the areas far away from the urban centres of the Punjab, we need information about locations and size of communities in those regions. Since this information is not readily available, we have employed deep learning to detect house-like-structures on the freely available low resolution satellite imagery (visible spectrum) of the Cholistan Desert.

Read More

Is Economics From Afar Domain Generalizable?

To overcome the challenges of lack of information and regular surveys, machine-learning based satellite imagery algorithms have been proposed to predict the economic indicators of a region. These indicators (ranging from poverty estimation, slum detection) not only help government design policies but are vital tools for the businesses to understand their customers, design their business strategy and evaluate their business model. Unfortunately, these algorithms’ predictions are not interpretable and being based on homogeneity, they fail to generalize over the regions. Instead of popular black-box techniques, we plan to create an interpretable economic well-being analysis using satellite imagery and geo-spatial datasets and solve the domain adaptation problem by constraining over the interpretation rather than just generic image or feature level adaptation.