Citation

Ghorbanzadeh, O., Xu, Y., Zhao, H., Wang, J., Zhong, Y., Zhao, D., Zang, Q., Wang, S., Zhang, F., Shi, Y., Zhu, X.X., Bai, L., Li, W., Peng, W., Ghamisi, P. (2022). The Outcome of the 2022 Landslide4Sense Competition: Advanced Landslide Detection From Multisource Satellite Imagery. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 15: 9927-9942. Link to paper

Abstract

This paper presents the scientific outcomes of the 2022 Landslide4Sense (L4S) competition organized by the Institute of Advanced Research in Artificial Intelligence, focusing on automated landslide detection from global satellite imagery. The competition emphasized deep learning models for semantic segmentation tasks using a benchmark dataset combining multisource satellite observations. Winning solutions employed state-of-the-art models including the Swin Transformer, SegFormer, and U-Net, alongside advanced techniques such as hard example mining, self-training, and mix-up data augmentation. The benchmark dataset and results are made available on the Future Development Leaderboard for ongoing evaluation and community contribution. This competition advanced the state-of-the-art in automated landslide mapping by bringing together global expertise and establishing standardized evaluation protocols for comparing deep learning approaches. The outcomes provide valuable insights into effective architectures and training strategies for operational landslide detection systems.

AI Visual Interpretation (Beta)