Content Details | NUML Online Research Repository

Content Details

Back to Department Listing

Title AN IMPROVED CLASSIFICATION OF UNDERWATER SHIP-ENGINE AUDIOS USING SIAMESE NETWORK

Author(s) Maryam Sajid

Abstract Developing a reliable ship classification system using underwater acoustics is crucial due to limited labeled data, dynamic underwater conditions, and noise interference, reducing reliance on human sonar operators vulnerable to weather and fatigue. Improving underwater acoustic target classification requires addressing shortcomings in feature extraction, dataset availability, feature diversity, and classifier selection, but integrating multiple techniques must balance gains against increased costs, time, and system complexity. This study proposes a novel feature extraction technique that reduces the computational cost, complexity and increases the robustness of model. In this proposed technique, audios are segmented into chunks and spectrograms are calculated for them. The dataset is arranged in the form of triplets that are fed into the siamese network that are based on triplet loss, generates feature vectors. The goal of the siamese network is to learn an embedding space in which similar classes are grouped together and dissimilar classes are further separated. These extracted features may be fed into a classifier. Classifier will then classify the correct classes on the basis of given results. The model’s performance is evaluated on Shipears dataset. Furthermore, accuracy, precision, recall, f1-score and ROC curve are used to evaluate the performance of popular classifiers, k-NN, SVM, RF, DT. Overall accuracy of our model reaches 96.4167% which reduces the complexity.

Type Thesis/Dissertation

Faculty Engineering and Computer Science

Department Computer Science

Language English

Publication Date

Subject

Publisher

Contributor(s)

Format

Identifier

Source

Relation

Coverage

Rights