Skip to main content Skip to secondary navigation

Assessing brick kilns number, location and use in Bangladesh

Main content start
Map of the locations of brick kilns generated by the machine learning model
Air Pollution

Landscape of brick manufacturing in Bangladesh: a scalable approach to deep learning for sustainable development

 

Objective

Leverage advances in machine learning and improvements in satellite imagery to develop a reproducible and automated pipeline that locates and produces a database all of the traditional brick kilns in Bangladesh.

 

Rationale

Brick production is central to construction in Bangladesh, resulting in the rapid expansion of brick kilns throughout the country. However, there is no accurate accounting of how many kilns are operating as many are either not counted because they violate various regulations or are never registered, enabling further violations of national regulations. We will use this data to catalyze a discussion among brick kiln owners, government regulators, researchers and civil society. The long-term aim is to motivate progress toward a manufacturing system that generates less environmental and health harm.

 

Project Dates

2016-2020

 

Stage of Work

We have currently developed and tested a convolutional neural network that works well at identifying brick kilns in satellite imagery. We are conducting further refinements to the model, as well as developing approaches to isolate where within a given image a kiln is located in order to determine the approximate geographic locations of all kilns.

 

To Learn More About This Work

A Better Brick: Solving an Airborne Health Threat (Stanford Woods Institute for the Environment)

 

People

Primary Contact:  Nina Brooks

Stanford University

Stephen Luby

Nina Brooks

Jihyeon Lee

Fahim Tajwar

 

Funding

Stanford Woods Institute for the Environment: Environmental Venture Projects (EVP)