GASP - Gauging Air Status With Photography

Problem & Motivation

Air pollution is modern day fact of life. It is everywhere from the urban landscapes of Los Angeles to the remote villages of India. It is a ubiquitous challenge with tremendous health impacts. This is particularly true for the 500 million people around the world that suffer from chronic respiratory disease, which the third leading cause of death globally. For people struggling with this deadly ailment, air pollution is the second leading risk factor.

Unfortunately, many people living in developing nations or other remote, sparsely populated areas don’t have good access to reliable information about the air that they breathe. GASP seeks to solve this problem by utilizing a resource that many already do have—their smartphone. By using ML image processing techniques GASP provides users with a quick, inexpensive app to determine the current air quality classification. GASP also allows users to determine which groups should be concerned about the current air quality and what actions should be taken to protect their health.

Data Source

We have utilized three three existing datasets comprising over 14,000 images taken across India, Nepal, and Bangladesh. Each image is paired with one of seven different air quality labels ranging from “Good” to “Severe”, indicating the air quality represented by each image.

Our primary dataset was published by Utomo, et al. from the Graduate Institute of Ambient Intelligence and Smart Systems, National Chung Cheng University, Taiwan. It includes detailed information including image location, date, time, air quality classification, and various pollutant concentrations. It comprises over 12,000 of the 14,000 total images used in our analysis and is available on GitHub at github.com/ICCC-Platform/Air-Pollution-Image-Dataset-From-India-and-Nepal.

Data Science Approach

Data Preprocessing

We have taken three primary preprocessing steps:

(1) Resizing: adjusted to a standard size (224, 224, 3)

(2) Downsampling: eliminating some images that were taken at the same place and time

(3) Image augmentation: translation, rotation, or crop to create new data and correct for class imbalances in training data

Models

We have used two primary families of models:

(1) Pretrained CNN models including VGG16 and Efficient Net

(2) Pretrained Vision Transformer models including ViT by Google

Evaluation

We have assessed our models using a number of accuracy metrics including, overall accuracy, mean squared error, mean absolute error, and S-Corr. Our primary means of assessing model performance has been confusion matrices.

Given our model is intended to be used to protect health and safety, we have judged that underestimating the pollution level is more harmful than overestimating. Consequently we have chosen to assign a larger weight to underestimations than overestimations. This causes the model to error high more often than low, which is the safer option for users.

Final overall accuracy is XX%, more than twice what we were able to achieve with our baseline CNN model.

Key Learnings & Impact

Traditional air quality monitoring requires costly equipment that measures a number of different pollutants including: particulate matter, ozone, nitrogen dioxide, carbon monoxide, and sulfur dioxide. Given that a number of these pollutants are invisible, attempting to accurately classify air quality with photography equipment alone is a challenging task. As such, our accuracy is below what we would ideally like to be able to provide. We do believe however, that we have successfully demonstrated the potential of such an application. With a larger dataset taken from a wider variety of geographic locations, we believe significantly higher accuracy is achievable.

Acknowledgements

We would like to thank our Capstone instructors, Zona Kostic and Todd Holloway for their guidance and technical recommendations throughout the semester.