ROSAR: An Adversarial Re-Training Framework for Robust Side-Scan Sonar Object Detection
This work addresses robustness challenges for autonomous underwater vehicle applications using side-scan sonar, presenting an incremental improvement over prior knowledge distillation methods.
The paper tackles the problem of improving robustness of deep learning object detection models for side-scan sonar images by introducing ROSAR, a framework that integrates knowledge distillation with adversarial retraining, resulting in up to 1.85% enhancement in model robustness under specific conditions.
This paper introduces ROSAR, a novel framework enhancing the robustness of deep learning object detection models tailored for side-scan sonar (SSS) images, generated by autonomous underwater vehicles using sonar sensors. By extending our prior work on knowledge distillation (KD), this framework integrates KD with adversarial retraining to address the dual challenges of model efficiency and robustness against SSS noises. We introduce three novel, publicly available SSS datasets, capturing different sonar setups and noise conditions. We propose and formalize two SSS safety properties and utilize them to generate adversarial datasets for retraining. Through a comparative analysis of projected gradient descent (PGD) and patch-based adversarial attacks, ROSAR demonstrates significant improvements in model robustness and detection accuracy under SSS-specific conditions, enhancing the model's robustness by up to 1.85%. ROSAR is available at https://github.com/remaro-network/ROSAR-framework.