Using deep learning to track time × frequency whistle contours of toothed whales without human-annotated training data
Advanced Search
Select up to three search categories and corresponding keywords using the fields to the right. Refer to the Help section for more detailed instructions.

Search our Collections & Repository

For very narrow results

When looking for a specific result

Best used for discovery & interchangable words

Recommended to be used in conjunction with other fields

Dates

to

Document Data
Library
People
Clear All
Clear All

For additional assistance using the Custom Query please check out our Help Page

The NOAA IR serves as an archival repository of NOAA-published products including scientific findings, journal articles, guidelines, recommendations, or other information authored or co-authored by NOAA or funded partners. As a repository, the NOAA IR retains documents in their original published format to ensure public access to scientific information.
i

Using deep learning to track time × frequency whistle contours of toothed whales without human-annotated training data

Filetype[PDF-6.29 MB]


Select the Download button to view the document
This document is over 5mb in size and cannot be previewed
Details You May Also Like

Details:

  • Journal Title:
    The Journal of the Acoustical Society of America
  • Personal Author:
  • NOAA Program & Office:
    CIMAR (Cooperative Institute for Marine and Atmospheric Research)
  • Description:
    Many odontocetes produce whistles that feature characteristic contour shapes in spectrogram representations of their calls. Automatically extracting the time × frequency tracks of whistle contours has numerous subsequent applications, including species classification, identification, and density estimation. Deep-learning-based methods, which train models using analyst-annotated whistles, offer a promising way to reliably extract whistle contours. However, the application of such methods can be limited by the significant amount of time and labor required for analyst annotation. To overcome this challenge, a technique that learns from automatically generated pseudo-labels has been developed. These annotations are less accurate than those generated by human analysts but more cost-effective to generate. It is shown that standard training methods do not learn effective models from these pseudo-labels. An improved loss function designed to compensate for pseudo-label error that significantly increases whistle extraction performance is introduced. The experiments show that the developed technique performs well when trained with pseudo-labels generated by two different algorithms. Models trained with the generated pseudo-labels can extract whistles with an F1-score (the harmonic mean of precision and recall) of 86.31% and 87.2% for the two sets of pseudo-labels that are considered. This performance is competitive with a model trained with 12 539 expert-annotated whistles (F1-score of 87.47%).
  • Keywords:
  • Source:
    The Journal of the Acoustical Society of America, 154(1), 502-517
  • DOI:
  • ISSN:
    0001-4966
  • Format:
    PDF
  • Publisher:
    Acoustical Society of America (ASA)
  • Document Type:
    Journal Article
  • Rights Information:
    Other
  • Compliance:
    Submitted
  • Main Document Checksum:
  • Download URL:
  • File Type:

Supporting Files

  • No Additional Files
More +

You May Also Like

Checkout today's featured content at repository.library.noaa.gov

Version 3.27.1