Comparison of Deep Neural Network Architecture for Iceberg Detection in Radar Images

The article considers methods for detecting icebergs on radar images (RLS), studies existing implementations of iceberg detection, including those using deep learning. A search for Sentinel-1 satellite imagery data was performed, and preliminary image processing was performed. Existing ar-chitectures of neural network models for detecting objects on an image were studied, relevant architectures were selected, and neural network models were trained. A comparison of the trained models was performed, and the highest quality was selected. A program for interaction with the trained neural network was developed.