Landslide4Sense 2022: Multi-Source Remote Sensing for Landslide Detection

Landslide4Sense 2022 provides multi-source remote sensing imagery and landslide labels for a deep learning competition organized by IARAI in 2022. The data is designed for natural hazard detection and includes baseline PyTorch code for model development.

Use Cases

• Training semantic segmentation models using remote sensing image patches to delineate landslide boundaries
• Benchmarking PyTorch-based deep learning models for natural hazard classification
• Evaluating multi-source data fusion techniques using the provided satellite imagery layers

Strengths

• MIT licensed for open research
• Competition-grade annotations
• Includes baseline PyTorch implementation

Limitations

• Unknown record count and specific sensor details in the provided metadata
• Geographic bias likely limited to specific landslide event locations
• Temporal staleness as data was last updated in mid-2023

Provenance

Source

IARAI (Institute of Advanced Research in Artificial Intelligence)

Collection Method

Remote sensing imagery collection and expert annotation for competition use

Time Range

2022

Freshness

Last updated June 2023; static competition dataset.