Harbitz, A. 2007. Estimation of shrimp (Pandalus borealis) carapace length by image analysis. — ICES Journal of Marine Science, 64: 939–944.

The age determination of fish is fundamental to marine resource management. This task is commonly done by analysis of otoliths performed manually by human experts. Otolith images from Greenland halibut acquired by the Institute of Marine Research (Norway) were recently used to train a convolutional neural network (CNN) for automatically predicting fish age, opening the way for requiring less human effort and availability of expertise by means of deep learning (DL). In this study, we demonstrate that applying a CNN model trained on images from one lab (in Norway) does not lead to a suitable performance when predicting fish ages from otolith images from another lab (in Iceland) for the same species. This is due to a problem known as dataset shift, where the source data, i.e., the dataset the model was trained on have different characteristics from the dataset at test stage, here denoted as target data. We further demonstrate that we can handle this problem by using domain adaptation, such that an existing model trained in the source domain is adapted to perform well in the target domain, without requiring extra annotation effort. We investigate four different approaches: (i) simple adaptation via image standardization, (ii) adversarial generative adaptation, (iii) adversarial discriminative adaptation and (iv) self-supervised adaptation. The results show that the performance varies substantially between the methods, with adversarial discriminative and self-supervised adaptations being the best approaches. Without using a domain adaptation approach, the root mean squared error (RMSE) and coefficient of variation (CV) on the Icelandic dataset are as high as 5.12 years and 28.6%, respectively, whereas by using the self-supervised domain adaptation, the RMSE and CV are reduced to 1.94 years and 11.1%. We conclude that careful consideration must be given before DL-based predictors are applied to perform large scale inference. Despite that, domain adaptation is a promising solution for handling problems of dataset shift across image labs.

This paper focuses on artefacts that may corrupt stock discrimination by shape analysis of otolith contours, how one can examine if such artefacts are important, and how they can be avoided. The scope focuses on Fourier transforms of contour points, the linear Fisher discrimination technique, and success rates based on cross validation by the “leave one out at a time” technique. The “zero-score” technique is introduced as a tool to examine the importance of a possible artefact, based on the theoretical result that the probability of correct classification of any otolith from either of two identical groups is zero. If one of the identical groups is exposed to a possible influential factor, e.g. a different smoothing, a high classification rate will reveal that this factor is an important artefact. The concept of a “lasso contour” is introduced that drastically reduces the impact of smoothing and provides a non-concave shape that enables a one-dimensional representation of the contour without ambiguities. Results are illustrated by comparison between Greenland halibut (Reinhardtius hippoglossoides) otolith contours from southern Greenland and Northeast Arctic waters. The conclusion is that the probability of correct classification of locality based on the original contours is too optimistic (77–79%), while the scores based on lasso contours are insensitive to smoothing and still optimistically high (68–70%).

Harbitz, A., Ona, E., and Pennington, M. 2009. The use of an adaptive acoustic-survey design to estimate the abundance of highly skewed fish populations. – ICES Journal of Marine Science, 66: 1349–1354. The uncertainty (relative root-mean-square error) of abundance estimates based on a simple and easily implemented adaptive design for an acoustic survey is examined. The study is limited to surveys with parallel transects and independent observations with extremely skewed distributions. The adaptive approach defines a stratum for each conventional observation and how to take additional observations in that stratum when the acoustic density exceeds a predetermined threshold. The cost (sailing distance) of each added observation is about three times that of a conventional observation. This method was demonstrated using high-resolution transect data from a herring (Clupea harengus) acoustic survey conducted in 2006 off the north coast of Norway. The primary sampling unit for this winter survey was 5 nautical miles, and the usual distance between transects was 20 nautical miles. The results indicate that an adaptive survey design would substantially reduce the root-mean-square error of the abundance estimates compared with that of the conventional survey design.

The age structure of a fish population has important implications for recruitment processes and population fluctuations, and is key input to fisheries assessment models. The current method relies on manually reading age from otoliths, and the process is labor intensive and dependent on specialist expertise. Advances in machine learning have recently brought forth methods that have been remarkably successful in a variety of settings, with potential to automate analysis that previously required manual curation. Machine learning models have previously been successfully applied to object recognition and similar image analysis tasks. Here we investigate whether deep learning models can also be used for estimating the age of otoliths from images. We adapt a standard neural network model designed for object recognition to the task of estimating age from otolith images. The model is trained and validated on a large collection of images of Greenland halibut otoliths. We show that the model works well, and that its precision is comparable to and may even surpass that of human experts. Automating this analysis will help to improve consistency, lower cost, and increase scale of age prediction. Similar approaches can likely be used for otoliths from other species as well as for reading fish scales. The method is therefore an important step forward for improving the age structure estimates of fish populations.