Automatic Habitat Mapping using Convolutional Neural Networks

André Diegues, José Pinto and Pedro Ribeiro

2018

Abstract

In this paper we propose a novel technique for performing habitat mapping in submerged coastal areas using AUVs and Convolutional Neural Networks (CNNs). Our approach consists having multiple AUVs traveling close to the bottom to acquire geo-referenced photos. Habitats in the photos are (partially) identified by marine biologists, which makes a supervised learning approach possible. (preliminary version of abstract)

Digital Object Identifier (DOI)

doi 10.1109/AUV.2018.8729787

Publication in PDF format

pdf Download PDF

Journal/Conference/Book

IEEE OES Autonomous Underwater Vehicle Symposium

Reference (text)

André Diegues, José Pinto and Pedro Ribeiro. Automatic Habitat Mapping using Convolutional Neural Networks. Proceedings of the IEEE OES Autonomous Underwater Vehicle Symposium (AUV), IEEE, Porto, Portugal, November, 2018.

Bibtex

@inproceedings{ribeiro-AUV2018,
  author = {André Diegues and  José Pinto and Pedro Ribeiro},
  title = {Automatic Habitat Mapping using Convolutional Neural Networks},
  doi = {10.1109/AUV.2018.8729787},
  booktitle = {IEEE OES Autonomous Underwater Vehicle Symposium},
  publisher = {IEEE},
  month = {November},
  year = {2018}
}