Machine Learning Applications in Fish Age Estimation

Accurate determination of fish age is fundamental to fisheries management, ecosystem modelling and conservation. Traditional methods rely on manual interpretation of otoliths or scales, a time-consuming process prone to reader bias and inconsistency. Recent advances in machine learning have enabled automated, scalable and more objective assessment of age structures by analysing high-resolution images. Convolutional neural networks (CNNs) have been employed to recognise growth rings in otoliths with precision comparable to human experts, while techniques such as transfer learning, few-shot learning and domain adaptation reduce the need for large annotated datasets. Object-detection algorithms and semantic segmentation models have further refined ring counting by isolating regions of interest and enhancing feature continuity. These developments promise faster throughput, standardised outputs and broader coverage of species and geographies, supporting adaptive management and monitoring in a changing climate.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Machine Learning Applications in Fish Age Estimation publication trend

The graph below shows the total number of articles in machine learning applications in fish age estimation across all publications each year (not limited to Nature Index journals).

Technical terms

Otolith: A calcified structure in the inner ear of fish that forms annual growth rings, used as the primary biological record for age estimation.

Convolutional Neural Network (CNN): A class of deep learning model designed to process grid-structured data such as images, using convolutional layers to detect hierarchical features.

Transfer Learning: The practice of fine-tuning a model pre-trained on a large, general dataset to perform a related task with limited domain-specific data.

Few-Shot Learning: A machine learning approach that enables models to generalise to new classes or tasks from only a few annotated examples.

Domain Adaptation: Techniques that adjust a model trained on one dataset (source domain) to maintain performance on a different but related dataset (target domain) without extensive new annotations.

Semantic Segmentation: The process of classifying each pixel in an image into meaningful categories, here used to delineate growth zones in otolith images.

Object Detection: The identification and localisation of distinct regions or objects within an image, employed to isolate otolith reading axes or annular features.

References

1. Automatic interpretation of otoliths using deep learning. PLOS ONE (2018).
2. Explaining decisions of deep neural networks used for fish age prediction. PLOS ONE (2020).
3. Automatic Fish Age Determination across Different Otolith Image Labs Using Domain Adaptation. Fishes (2022).
4. Age prediction by deep learning applied to Greenland halibut (Reinhardtius hippoglossoides) otolith images. PLOS ONE (2022).
5. Otolith age determination with a simple computer vision based few-shot learning method. Ecological Informatics (2023).
6. Fish age reading using deep learning methods for object-detection and segmentation. ICES Journal of Marine Science (2024).