@article{fdi:010081020,
  title = {{C}omplex data labeling with deep learning methods : lessons from fisheries acoustics},
  author = {{S}arr, {J}. {M}. {A}. and {B}rochier, {T}imoth{\'e}e and {B}rehmer, {P}atrice and {P}errot, {Y}annick and {B}ah, {A}. and {S}arre, {A}. and {J}eyid, {M}. {A}. and {S}idibeh, {M}. and {E}l {A}youbi, {S}.},
  editor = {},
  language = {{ENG}},
  abstract = {{Q}uantitative and qualitative analysis of acoustic backscattered signals from the seabed bottom to the sea surface is used worldwide for fish stocks assessment and marine ecosystem monitoring. {H}uge amounts of raw data are collected yet require tedious expert labeling. {T}his paper focuses on a case study where the ground truth labels are non-obvious: echograms labeling, which is time-consuming and critical for the quality of fisheries and ecological analysis. {W}e investigate how these tasks can benefit from supervised learning algorithms and demonstrate that convolutional neural networks trained with non-stationary datasets can be used to stress parts of a new dataset needing human expert correction. {F}urther development of this approach paves the way toward a standardization of the labeling process in fisheries acoustics and is a good case study for non-obvious data labeling processes.},
  keywords = {{F}isheries acoustics ; {M}achine learning ; {N}eural network ; {A}ctive acoustics ; {L}abeling process ; {B}ottom correction},
  booktitle = {},
  journal = {{I}sa {T}ransactions},
  volume = {109},
  numero = {},
  pages = {113--125},
  ISSN = {0019-0578},
  year = {2021},
  DOI = {10.1016/j.isatra.2020.09.018},
  URL = {https://www.documentation.ird.fr/hor/fdi:010081020},
}