Robust underwater sediment estimation using frequency analysis of environmental sounds.

In this study, the accuracy may be extremely reduced when the sea area is different in the classification by CNN based on the seabed image taken so as to crawl on the seabed. Therefore, the purpose is to improve the accuracy by utilizing the acoustic information accompanying it. In the classification by sound, the classification by CNN using spectrogram images can be mentioned as the classification of environmental sounds. It can be said that the acoustic characteristics of each sediment change in the same way as the sea area changes, and we will aim to improve the accuracy in the future by paying attention to this.