Marine Biodiversity and Ecosystem Functioning
EU Network of Excellence

 
Main Menu

· Home
· Contacts
· Data Systems
· Documents
· FAQ
· Links
· MarBEF Open Archive
· Network Description
· Outreach
· Photo Gallery
· Quality Assurance
· Register of Resources
· Research Projects
· Rules and Guidelines
· Training
· Wiki
· Worldconference

 

Register of Resources (RoR)

 People  |  Datasets  |  Literature  |  Institutes  |  Projects 

[ report an error in this record ]basket (1): add | show Print this page

one publication added to basket [362027]
Autonomous temporal pseudo-labeling for fish detection
Veiga, R.J.M.; Ochoa, I.E.; Belackova, A.; Bentes, L.; Silva, J.P.; Semião, J.; Rodrigues, J.M.F. (2022). Autonomous temporal pseudo-labeling for fish detection. Applied Sciences-Basel 12(12): 5910. https://dx.doi.org/10.3390/app12125910
In: Applied Sciences-Basel. MDPI: Basel. e-ISSN 2076-3417
Peer reviewed article  

Available in  Authors 

Keywords
    Pisces [WoRMS]
    Marine/Coastal
Author keywords
    environmental monitoring; marine fishes; object detection; fish detection; pseudo-labeling; underwater video; deep learning

Authors  Top 
  • Veiga, R.J.M.
  • Ochoa, I.E.
  • Belackova, A.
  • Bentes, L.
  • Silva, J.P.
  • Semião, J.
  • Rodrigues, J.M.F.

Abstract
    The first major step in training an object detection model to different classes from the available datasets is the gathering of meaningful and properly annotated data. This recurring task will determine the length of any project, and, more importantly, the quality of the resulting models. This obstacle is amplified when the data available for the new classes are scarce or incompatible, as in the case of fish detection in the open sea. This issue was tackled using a mixed and reversed approach: a network is initiated with a noisy dataset of the same species as our classes (fish), although in different scenarios and conditions (fish from Australian marine fauna), and we gathered the target footage (fish from Portuguese marine fauna; Atlantic Ocean) for the application without annotations. Using the temporal information of the detected objects and augmented techniques during later training, it was possible to generate highly accurate labels from our targeted footage. Furthermore, the data selection method retained the samples of each unique situation, filtering repetitive data, which would bias the training process. The obtained results validate the proposed method of automating the labeling processing, resorting directly to the final application as the source of training data. The presented method achieved a mean average precision of 93.11% on our own data, and 73.61% on unseen data, an increase of 24.65% and 25.53% over the baseline of the noisy dataset, respectively.

All data in the Integrated Marine Information System (IMIS) is subject to the VLIZ privacy policy Top | Authors 


If any information here appears to be incorrect, please contact us
Back to Register of Resources
 
Quick links

MarBEF WIKI

Erasmus Mundus Master of Science in Marine Biodiversity and Conservation (EMBC)
Outreach

Science
Responsive Mode Programme (RMP) - Marie Nordstrom, copyright Aspden Rebecca

WoRMS
part of WoRMS logo

ERMS 2.0
Epinephelus marginatus Picture: JG Harmelin

EurOBIS

Geographic System

Datasets

 


Web site hosted and maintained by Flanders Marine Institute (VLIZ) - Contact data-at-marbef.org