BrazilDam Dataset

 BrazilDAM Dataset

BrazilDAM is a multi sensor and multitemporal dataset that consists of multispectral images of ore tailings dams throughout Brazil. Landsat 8 and Sentinel 2 satellites that capture multispectral images over the years 2016, 2017, 2018 and 2019 were used. The dataset contains samples collected in different regions, which increases the diversity and representativeness of the characteristics of the dams.

 

Coordinates data came from the Brazilian National Mining Agency that's made available on its website a georeferenced database describing all officially registered tailings dams in the country. The database contains some identification information about the dams and their owners and spatial coordinates of a point indicating each dam’s location. There is also some technical information like the main ore, height, volume, constructive method, risk category, potential damage associated with the dam.

2016

2017

2018

2019

LANDSAT

SENTINEL

Statistics

Results of methods and tools evaluated against the RCPD dataset:

Satellite Landsat 8 Sentinel 2
Year Dam Not Dam Dam Not Dam
2016 761 1143 769 1156
2017 769 1155 769 1156
2018 769 1156 769 1156
2019 756 1130 769 1156

Citation

The dataset is described in the paper BrazilDAM: A Benchmark dataset for Tailings Dam Detection

If you mention the BrazilDAM dataset, please use the following reference:

 
@inproceedings{
  ferreira2020brazildam,
  title={BrazilDAM: A Benchmark dataset for Tailings Dam Detection},
  author={Ferreira, E., Brito, M., Alvim, M. S., Balaniuk, R., dos Santos, J.},
  booktitle={Latin American GRSS & ISPRS Remote Sensing Conference 2020, Santigo, Chile},
  year={2020},
  organization={IEEE}
}

Download Dataset/Code