Ship Classification in SAR Images via Densely Connected Triplet CNNs Integrating Fisher Discrimination Regularized Metric Learning

This paper focuses on ship classification from Synthetic Aperture Radar (SAR) images using a Triplet Convolutional Neural Network (CNN) integrated with Fisher Discrimination Regularized Metric Learning. The system involves dataset creation, triplet sampling, model training, and ship type prediction (Cargo, Passenger, Dredging, Tanker). A GUI built with Tkinter allows users to upload SAR images and view the prediction results.