• Aucun résultat trouvé

A deep learning algorithm for automatic identification of coral reef fish species on images

N/A
N/A
Info
Télécharger
Protected

Academic year: 2021

Partager "A deep learning algorithm for automatic identification of coral reef fish species on images"

Copied!
2
0
0

Chargement.... (Voir le texte intégral maintenant)

Texte intégral

(1)

HAL Id: lirmm-01883997

https://hal-lirmm.ccsd.cnrs.fr/lirmm-01883997

Submitted on 29 Sep 2018

HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

A deep learning algorithm for automatic identification of coral reef fish species on images

Sebastien Villon, David Mouillot, Marc Chaumont, Emily Darling, Gérard Subsol, Thomas Claverie, Sébastien Villéger

To cite this version:

Sebastien Villon, David Mouillot, Marc Chaumont, Emily Darling, Gérard Subsol, et al.. A deep learning algorithm for automatic identification of coral reef fish species on images. WCMB: World Conference on Marine Biodiversity, May 2018, Montréal, Canada. �lirmm-01883997�

(2)

A deep learning algorithm for automatic identification of coral reef fish species on images

Sébastien Villon

c,a

, David Mouillot

c

, Marc Chaumont

b,a

, Emily Darling

d

, Gérard Subsol

a

, Thomas Claverie

c,e

,Sébastien Villéger

c

sebastien.villon@lirmm.fr

a LIRMM, University of Montpellier/CNRS, France b University of Nîmes, France

c MARBEC, IRD/Ifremer/University of Montpellier/CNRS, France d WCS World Life Conservation Society, New-York, U.S.A.

e CUFR Mayotte, France

One of the current challenges of marine ecology is to monitor biodiversity accurately and efficiently at large spatial scale and at high frequency. Therefore, underwater surveys based on video are increasingly used as an alternative to reef fish count by divers. However, analyzing videos imped the process of fast quantification because it remains time-consuming. There is thus an urgent need for automatic identification of coral reef fish species in underwater images. We designed a convolutional neural network and trained it using a learning database containing more than 90,000 fish images of 20 different common species on Mayotte coral reef. In experiments based on more than 4,400 test images from different acquisition campaigns, this algorithm reached 94% of accurate recognition. In particular, the algorithm was able to recognize a fish even when it was partially hidden behind corals or other fish. We also tested the effectiveness of our algorithm by comparing it to human specialists in terms of speed and correct identification. The algorithm outperforms human experts by more than 5% (95.7% against 89.3%) and it was around 100 times faster than human (processing an image in 0.06 seconds).

Keywords: Deep Learning, fishes, Identification, Images processing

Références

Télécharger maintenant ( PDF - 2 Page - 137.72 KB )

Documents relatifs

FISH SPECIES IDENTIFICATION USING IMAGE ANALYSIS OF ECHO-SOUNDER IMAGES

Using a 120 KHz transducer Rose and Leggett measured the relationship between fish length and target strength for each species and their results were in agreement with those of

A new method to control error rates in automated species identification with deep learning algorithms Sébastien Villon

processing with the threshold selected by optimizing the Misclassification rate (threshold tuned and tested on the same dataset).. 4: Rates of unsure, correct, and

Classification d’objets urbains à partir de données LiDAR 3D terrestre par Deep-Learning Younes ZEGAOUI ¹² Marc CHAUMONT ²³ Gérard SUBSOL ² Philippe BORIANNE ⁴ Mustapha DERRAS ¹

1xK input points cloud 512x3 aligned points 512x3 MLP (64, 64) point features 512x64T-Net T-Net aligned point features 512x64 MLP (64,128, 1024). ● N’utilise pas de structure

A Deep Learning method for accurate and fast identification of coral reef fishes in underwater images Sébastien Villon

average identification rate since fish individuals are seen in all positions. The second training dataset T2 contained fish thumbnails from T1 plus “part of fish” thumbnails.

A deep learning algorithm for automatic identification of coral reef fish species on images

We designed a convolutional neural network and trained it using a learning database containing more than 90,000 fish images of 20 different common species on Mayotte coral reef.

images with Deep Learning algorithms

[1] Sebastien Villon, Marc Chaumont, Gerard Subsol, Sebastien Villeger, Thomas Claverie, David Mouillot, "Coral reef fish detection and recognition in underwater videos

A Deep learning method for accurate and fast identification of coral reef fishes in underwater images

“Environment”. Environment thumbnails were extracted at random in portion of frames where no fish was detected. We hypothesized that such a procedure can help distinguishing

Exposure to agricultural pesticide impairs visual lateralization in a larval coral reef fish

Using a range of aquarium based (Fig. 1) and in situ experiments we tested whether larvae exhibit a naïve preference for their left or right eye when inspecting both positive