Fig 1.
Study area in rural-urban Bengaluru, India, and training/ validation data boundary.
Map was created using ArcGIS (version: 3.5) from ESRI (http://www.arcgis.com/) under the institutional ESRI site license of the University of Kassel. Basemap satellite image accessed from World Imagery ESRTI Tile Layer. Credits: Esri, Earthstar Geographics, TomTom, Garmin, FAO, NOAA, USGS, © OpenStreetMap contributors, and the GIS User Community.
Fig 2.
Aerial picture of Koira granite quarry in Bengaluru, India, taken by the authors in July 2023.
Table 1.
Characteristics of used spectral bands of Sentinel-2 images for the study of granite mines around Bengaluru, South India.
Fig 3.
Mining site detection workflow based on deep learning for the study of granite mines around Bengaluru, South India.
Table 2.
Segmentation Accuracy Metrics (TP = True Positive; FP = False Positive; FN = False Negative) for the study of granite mines around Bengaluru, South India.
Table 3.
In-area performance of five deep learning architectures for granite quarry segmentation in Bengaluru, South India.
Table 4.
External audit (n = 400): Accuracy ± 95% Wilson CI, Precision, Recall, F1 ± 95% bootstrap CI, and IoU.
Fig 4.
Generalization gap: bar pairs per model (in-area IoU vs external IoU).
Fig 5.
Qualitative performance in the external validation AOI (unseen during training) of PSPNet, DeepLabV3+ , U-Net, FCN, and EMANet of four granite quarries around Bengaluru, South India.
Map was created using ArcGIS (version: 3.5) under the institutional ESRI site license of the University of Kassel, Germany (http://www.arcgis.com/). Basemap satellite image accessed from World Imagery ESRTI Tile Layer. Credits: Esri, Maxar, Earthstar Geographics, TomTom, Garmin, FAO, NOAA, USGS, © OpenStreetMap contributors, and the GIS User Community.
Fig 6.
Micro-AOI TP/FP/FN overlays + the 3-panel failure cases for PSPNet model.
Map was created using ArcGIS (version: 3.5) under the institutional ESRI site license of the University of Kassel, Germany (http://www.arcgis.com/). Basemap satellite image accessed from World Imagery ESRTI Tile Layer. Credits: Esri, Maxar, Earthstar Geographics, TomTom, Garmin, FAO, NOAA, USGS, © OpenStreetMap contributors, and the GIS User Community.
Fig 7.
(A) Distribution of active and abandoned granite quarry in Bengaluru (South India) of 2024.
(B) Heatmap of granite quarry density. It was created using ArcGIS (version: 3.5) under the institutional ESRI site license of the University of Kassel, Germany.
Fig 8.
High-resolution 3D mesh of the Prasannacharipalya quarry near Bengaluru (South India), generated with Pix4Dmapper version 4.9.0 shows excavation areas and granite formations (A) Textured shader (B) Altitude shader (Red, Green, Blue).
To this end authors’ drone images taken in July 2023 were mosaicked.
Fig 9.
(A) High resolution mosaic of authors’ drone images taken in July 2023 showing the border of volumetric analysis of the Prasannacharipalya granite quarry near Bengaluru, South India and (B) textured 3D mesh showcasing the pre-quarry and post-quarry elevation.
Table 5.
Volumes and uncertainties associated with surface lowering at extraction sites of granite mines around Bengaluru, South India.
Table 6.
Results of the Monte-Carlo experiment (lowering-only) at 1 m grid.