Conditional Random Fields for Improving Deep Learning-Based Glacier Calving Front Delineations

Advancements in Deep Learning have enabled the automated identification of glacier calving fronts in satellite imagery. This study improves the accuracy of this process by incorporating a Conditional Random Field (CRF) into the post-processing of the neural network’s predictions. Experiments using the CaFFe dataset showed a 27-meter improvement in mean distance error. The code is available at