Matching human-level performance with Convolutional Neural Networks on delineation of glacier calving fronts

Outline of processing pipeline from Mohajerani et al. [2019] Remote Sensing.
doi.org/10.3390/rs11010074d

Our new paper on the use of convolutional neural networks to delineate glacier calving fronts in Greenland came out in Remote Sensing. This allows fast and efficient delineation of calving fronts across the ice sheets, allowing a better understanding of glacier dynamics. While previous studies relied on manual delineation to study a few glaciers at a time, data-science approaches, and specifically the use of machine learning techniques in the geosciences open up new opportunities from eliminating the bottleneck of limited interpretable data. We achieve human-level performance in detecting the fronts! Check out the paper.

Mohajerani, Y.,Wood, M., Velicogna, I., Rignot, E. “Detection of Glacier Calving Margins with Convolutional Neural Networks: A Case Study.” Remote Sensing 11.1 (2019): 74. http://doi.org/doi:10.3390/rs11010074

One Reply to “Matching human-level performance with Convolutional Neural Networks on delineation of glacier calving fronts”

Leave a Reply

Your email address will not be published. Required fields are marked *