Best Practices in Convolutional Networks for Forward-Looking Sonar Image Recognition

TL;DR

This paper evaluates key design choices in CNNs for sonar image recognition, demonstrating effective transfer learning, the impact of dataset size, and the benefits of specific training techniques for small datasets.

Contribution

It provides a systematic analysis of transfer learning, image size, and dataset size effects on CNN performance in sonar image recognition, offering practical guidelines.

Findings

Transfer learning with SVM performs well even with disjoint classes.

High accuracy achievable with small datasets and small images using ADAM and Batch Normalization.

At least 50 samples per class needed for 90% accuracy.

Abstract

Convolutional Neural Networks (CNN) have revolutionized perception for color images, and their application to sonar images has also obtained good results. But in general CNNs are difficult to train without a large dataset, need manual tuning of a considerable number of hyperparameters, and require many careful decisions by a designer. In this work, we evaluate three common decisions that need to be made by a CNN designer, namely the performance of transfer learning, the effect of object/image size and the relation between training set size. We evaluate three CNN models, namely one based on LeNet, and two based on the Fire module from SqueezeNet. Our findings are: Transfer learning with an SVM works very well, even when the train and transfer sets have no classes in common, and high classification performance can be obtained even when the target dataset is small. The ADAM optimizer combined with Batch Normalization can make a high accuracy CNN classifier, even with small image sizes (16 pixels). At least 50 samples per class are required to obtain $90\%$ test accuracy, and using Dropout with a small dataset helps improve performance, but Batch Normalization is better when a large dataset is available.