Self-supervised Learning for Sonar Image Classification

TL;DR

This paper explores the use of self-supervised learning methods to improve sonar image classification for underwater robotics, demonstrating that these methods can achieve performance comparable to supervised learning in limited data scenarios.

Contribution

It evaluates three self-supervised learning techniques for sonar images and shows their effectiveness in transfer learning tasks without requiring labeled datasets.

Findings

Self-supervised pre-training matches supervised performance in few-shot learning.

All three methods show promising results on real sonar datasets.

Code and models are publicly available for further research.

Abstract

Self-supervised learning has proved to be a powerful approach to learn image representations without the need of large labeled datasets. For underwater robotics, it is of great interest to design computer vision algorithms to improve perception capabilities such as sonar image classification. Due to the confidential nature of sonar imaging and the difficulty to interpret sonar images, it is challenging to create public large labeled sonar datasets to train supervised learning algorithms. In this work, we investigate the potential of three self-supervised learning methods (RotNet, Denoising Autoencoders, and Jigsaw) to learn high-quality sonar image representation without the need of human labels. We present pre-training and transfer learning results on real-life sonar image datasets. Our results indicate that self-supervised pre-training yields classification performance comparable to supervised pre-training in a few-shot transfer learning setup across all three methods. Code and self-supervised pre-trained models are be available at https://github.com/agrija9/ssl-sonar-images