Automated fish age estimation from otolith images using statistical learning

Abstract

The acquisition of age and growth data is of key importance for fisheries research (assessment, marine ecology issues, etc.). Consequently, automating this task is of great interest. In this paper, we investigate the use of statistical learning techniques for fish age estimation. The core of this study lies in the definition of relevant image-related features. We rely on the computation of a 1D representation summing up the content of otolith images within a predefined area of interest. Features are then extracted from this non-stationary representation depicting the alternation of seasonal growth rings. Thus, fish age estimation can be viewed as a multi-class classification issue using statistical learning strategies. In particular, a procedure based on demodulation and remodulation of fish growth patterns is used to improve the generalization properties of the trained classifiers. The experimental evaluation is carried out over a dataset of 320 plaice otolith images from age groups 1-6. We analyze both, the performances of several statistical classifiers, namely SVMs (support vector machines) and neural networks, and the relevance of the proposed image-based feature sets. In addition, the combination of additional biological and shape features to the image-related ones is considered. We reach a rate of correct age estimation of 88% w.r.t. the expert ground truth. This demonstrates the relevance of the proposed approach for the automation of routine aging and for computer-assisted aging.

Statistical learning applied to computer-assisted fish age and growth estimation from otolith images

“…A variety of solutions can be designed to help in the acquisition of age and growth data (for instance, the automation of the acquisition of series of otolith images (Ogor and Fablet, 2004) or the storage and management of bases of images of interpreted calcified structures (CS) Ogor and Fablet, 2004)) or to automate this task thanks to the estimation of fish age and growth estimation from CS images (Fablet and Josse, 2005;Guillaud et al, 2002;Robertson and Morison, 1998;Troadec et al, 2000). The research effort has been mainly focused on the automation of the estimation of fish age and growth.…”

“…Hence, there is need for an improved automated extraction of the information conveyed by otolith images in terms of alternation of translucent and opaque rings. This issue was further investigated in (Fablet and Josse, 2005). A more robust information extraction stage was developed and successfully combined to statistical learning (namely support vector machines and neural networks) for fish age estimation.…”

“…They provide us with the methodological background to infer interpretation rules from training examples. Compared to (Fablet and Josse, 2005), we focus on kernel approaches (Scholkopf and Smola, 2002), especially kernel logistic regression (Zhu and Hastie, 2001) which provides us with probabilistic classification functions. This paper is organized as follows.…”

Otolith shape and size: The importance of age when determining indices for fish-stock separation

“…Machine learning methods have been used previously for otolith age classification (Fablet and Le Josse 2005;Bermejo et al 2007), and underlying methods have been used for stock separation with varying success.…”

Estimating age at maturation and energy-based life-history traits from individual growth trajectories with nonlinear mixed-effects models

“…Moreover, for many fish stocks, maturity data are too scarce to estimate a maturity ogive for each cohort, so that temporal variations in maturity cannot be accounted for. Coupled with automated otolith analyses (Fablet and Le Josse 2004), our method would complement a powerful toolbox to analyse large collections of otoliths and provide historical time series of maturity ogives, thereby improving the reconstruction of historical spawning stock biomass time series. The application is not restricted to fish, but can be applied to any other taxa with calcified structures allowing the back-calculation of individual growth trajectories.…”