Brick Kiln Dataset for Pakistan's IGP Region Using AI

Abstract

Brick kilns are a major source of air pollution in Pakistan, with many operating without regulation. A key challenge in Pakistan and across the Indo-Gangetic Plain is the limited air quality monitoring and lack of transparent data on pollution sources. To address this, we present a two-fold AI approach that combines low-resolution Sentinel-2 and high-resolution imagery to map brick kiln locations. Our process begins with a low-resolution analysis, followed by a post-processing step to reduce false positives, minimizing the need for extensive high-resolution imagery. This analysis initially identified 20,000 potential brick kilns, with high-resolution validation confirming around 11,000 kilns. The dataset also distinguishes between Fixed Chimney and Zigzag kilns, enabling more accurate pollution estimates for each type. Our approach demonstrates how combining satellite imagery with AI can effectively detect specific polluting sources. This dataset provides regulators with insights into brick kiln pollution, supporting interventions for unregistered kilns and actions during high pollution episodes.

Fig. 1

Study Area Map - (a) The Indian-Gangetic Plain (IGP) region, spanning across Bangladesh, India, and Pakistan (b) Focuses on the IGP region within Pakistan, which is the specific area of interest for kiln detection in this study.

Fig. 2

Image tiles illustrating different classes labelled for training our Random Forest Classifier.

Fig. 3

(a) The figure outlines the process of brick kiln detection in the Indus-Gangetic Plain using Sentinel-2 imagery. A 1 × 1 grid of the region is extracted, RGB bands are flattened into feature vectors, and a Random Forest classifier generates a mask identifying brick kiln locations. (b) The figure shows the visual representation of the steps involved in post-processing pipeline to accurately geolocate brick kilns in the image. (c) The figure illustrates the process of using the YOLOv8 model to detect brick kilns from Google Maps imagery, resulting in bounding boxes on the input image.

Fig. 4

Normalised Confusion Matrix for YOLOv8n.

Fig. 5

(a) and (b) represent two adjacent image tiles where the same kiln is partially visible in both.

Fig. 6

Overview of brick kiln distribution, types, and proximity to sensitive areas within IGP-Pakistan. (a) Distribution of brick kiln density across IGP-Pakistan, highlighting areas with high kiln concentrations and potential pollution hotspots. (b) Classification of kiln types (FCBK and zigzag) showing regional differences in kiln technology. (c) Percentage of schools, hospitals, and population density within 1 km of kilns, indicating exposure risks for nearby communities.