Deep Semantic Segmentation in an AUV for Online Posidonia Oceanica Meadows identification

TL;DR

This paper introduces a deep neural network for real-time, high-precision semantic segmentation of Posidonia oceanica meadows in underwater images, aiding ecological monitoring and conservation efforts.

Contribution

It presents a novel deep learning approach that outperforms existing methods and is integrated into an AUV for online mapping of underwater meadows.

Findings

Achieved 96.57% precision and 96.81% accuracy in segmentation.

Demonstrated real-time performance on an AUV.

Surpassed manual labeling reliability.

Abstract

Recent studies have shown evidence of a significant decline of the Posidonia oceanica (P.O.) meadows on a global scale. The monitoring and mapping of these meadows are fundamental tools for measuring their status. We present an approach based on a deep neural network to automatically perform a high-precision semantic segmentation of P.O. meadows in sea-floor images, offering several improvements over the state of the art techniques. Our network demonstrates outstanding performance over diverse test sets, reaching a precision of 96.57% and an accuracy of 96.81%, surpassing the reliability of labelling the images manually. Also, the network is implemented in an Autonomous Underwater Vehicle (AUV), performing an online P.O. segmentation, which will be used to generate real-time semantic coverage maps.