Introducing the Brand New QiandaoEar22 Dataset for Specific Ship Identification Using Ship-Radiated Noise

TL;DR

This work addresses the lack of publicly available, realistic multi-target underwater acoustic datasets by introducing QiandaoEar22, a multi-target dataset collected in Qiandaohu that includes 9h28m of ship-radiated noise and 21h58m of background noise across 20 target categories. The authors evaluate six deep learning classifiers across multiple feature representations to identify a specific ship type within multi-target signals, demonstrating that DenseNet with 2D features achieves the best performance, notably 97.78% accuracy for UUV identification using spectrum and MFCC features. The dataset provides a benchmark for underwater acoustic target detection (UATD) and recognition (UATR) and is expected to spur development of advanced methods for real-world underwater target identification. The work also details the data collection setup, labeling strategy, and baseline results to guide future research in underwater acoustics and related security applications.

Abstract

Target identification of ship-radiated noise is a crucial area in underwater target recognition. However, there is currently a lack of multi-target ship datasets that accurately represent real-world underwater acoustic conditions. To ntackle this issue, we release QiandaoEar22 \textemdash an underwater acoustic multi-target dataset, which can be download on https://ieee-dataport.org/documents/qiandaoear22. This dataset encompasses 9 hours and 28 minutes of real-world ship-radiated noise data and 21 hours and 58 minutes of background noise data. We demonstrate the availability of QiandaoEar22 by conducting an experiment of identifying specific ship from the multiple targets. Taking different features as the input and six deep learning networks as classifier, we evaluate the baseline performance of different methods. The experimental results reveal that identifying the specific target of UUV from others can achieve the optimal recognition accuracy of 97.78\%, and we find using spectrum and MFCC as feature inputs and DenseNet as the classifier can achieve better recognition performance. Our work not only establishes a benchmark for the dataset but helps the further development of innovative methods for the tasks of underwater acoustic target detection (UATD) and underwater acoustic target recognition(UATR).

Figures (3)

• Figure 1: Duration of the recorded data for different targets
• Figure 2: Target ships
• Figure 3: Average experimental results of deep learning methods for 2D feature classification