Hephaestus: A large scale multitask dataset towards InSAR understanding

Image | Volcano | Preparedness

Hephaestus is a volcano-disaster dataset for semantic segmentation of ground deformation. It contains 19,919 labeled and 110,573 unlabeled Sentinel-1 interferograms. It is introduced for binary classification.

  • ML task type: Binary classification
  • Data Source: Earth Observation Data and GeoSpatial Imagery (InSAR)
  • Size: 19,919 Labeled Images; 110,573 Unlabeled Images
  • Timespan: 2014 - 2021
  • Geographical Coverage: Global
  • Baseline Information
  • Evaluated on: SwinPL, ResNet18, ResNet18-MoCo for ground deformation
  • Metrics used: Accuracy, Precision, Recall, F-Score
  • Results as reported in original paper: Accuracy: 97.4%; Precision: 79.9%; Recall: 84.6% ; F-Score: 82.2% (for ResNet18 which has the highest accuracy)

Nikolaos Ioannis Bountos, Ioannis Papoutsis, Dimitrios Michail, Andreas Karavias, Panagiotis Elias, and Isaak Parcharidis. Hephaestus: A large scale multitask dataset towards insar understanding. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 1453–1462, 2022.