A Deep Learning Approach to Target Recognition in Side-Scan Sonar Imagery

Automatic target recognition using autonomous underwater vehicles and involving side-scan sonar imagery is of significant interest in both military and commercial applications. Adaptation of a state-of-the-art deep learning method for object detection in side-scan sonar imagery is described. Results of application of the method to available data support the approach as simple yet robust in detecting objects/anomalies along the seabed. A systematic procedure was employed in transfer learning a pre-trained convolutional neural network in order to learn the pixel-intensity based features of seafloor anomalies in sonar images. Using this process, newly trained convolutional neural network models were produced using relatively small training datasets and tested to show reasonably accurate anomaly detection and classification with little to no false alarms.