In this repository, we present two new deep learning architectures based on spatio-temporal modeling for underwater signal classification. And tested on two real datasets: DeepShip and ShipsEar.
- The first architecture is based on a static feature extraction (MFCC features).
- The second one, is an improvement of the first by adding a convolutional bloc to generate an artificial spectrum and makes the network takes the original signal as input directly.
model.py: DL models definition in Kerasparams.py: Hyperparametersfeatures.py: Dataset preparation and features extractioninference.py: Testing the modelfile.wav: wav file for test
| Model | Accuracy | Precision | Recall | F1-score |
|---|---|---|---|---|
| Hybrid model | 0.9878 | 0.9839 | 0.9949 | 0.9878 |
| End-to-end model | 0.9856 | 0.9730 | 0.9964 | 0.9843 |
| Model | Accuracy | Precision | Recall | F1-score |
|---|---|---|---|---|
| Hybrid model | 0.9727 | 0.9721 | 0.9731 | 0.9725 |
| End-to-end model | 0.9107 | 0.9010 | 0.9103 | 0.9097 |
If you use this code in your research, please consider citing this work via the following:
Plain Text
Z. Alouani, Y. Hmamouche, B. E. Khamlichi and A. E. F. Seghrouchni, "A Spatio-temporal Deep Learning Approach for Underwater Acoustic Signals Classification," 2022 18th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS), 2022, pp. 1-7, doi: 10.1109/AVSS56176.2022.9959247.
** APA **
Alouani, Z., Hmamouche, Y., El Khamlichi, B., & Seghrouchni, A. E. F. (2022, November). A Spatio-temporal Deep Learning Approach for Underwater Acoustic Signals Classification. In 2022 18th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS) (pp. 1-7). IEEE.
** ISO 690 **
ALOUANI, Zakaria, HMAMOUCHE, Youssef, EL KHAMLICHI, Btissam, et al. A Spatio-temporal Deep Learning Approach for Underwater Acoustic Signals Classification. In : 2022 18th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). IEEE, 2022. p. 1-7.
** MLA **
Alouani, Zakaria, et al. "A Spatio-temporal Deep Learning Approach for Underwater Acoustic Signals Classification." 2022 18th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS). IEEE, 2022.
BibTex
@INPROCEEDINGS{9959247, author={Alouani, Zakaria and Hmamouche, Youssef and Khamlichi, Btissam El and Seghrouchni, Amal El Fallah}, booktitle={2022 18th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS)}, title={A Spatio-temporal Deep Learning Approach for Underwater Acoustic Signals Classification}, year={2022}, volume={}, number={}, pages={1-7}, doi={10.1109/AVSS56176.2022.9959247}}
A Spatio-temporal Deep Learning Approach for Underwater Acoustic Signals Classification