Complex-Valued Neural Networks for Radar-Based Oil Spill Characterization: Development and Evaluation

Type of work:

Master Thesis

Summary:

This thesis investigates the use of complex-valued neural networks (CV-NNs) for radar-based oil spill characterization. The work will consider characterization performance, model complexity, and computational requirements to determine whether the use of CV-NNs provides a meaningful performance advantage. The analysis will be conducted using synthetic and real datasets. Depending on the results obtained, the feasibility of an FPGA implementation of the selected model is assessed, with particular emphasis on resource utilization and potential efficiency improvements.

Skills:

• Programming experience in Python
• Knowledge of deep learning
• Familiarity with PyTorch or TensorFlow
• Interest in radar signal processing and remote sensing
• Motivation to work on interdisciplinary topics combining radar, AI, and embedded systems

• Knowledge of FPGA design (using HLS).

   

Supervisor:

Dr.-Ing. Bilal Hammoud

Student:

Nacer Boudad

Year:

2026