shonsy / automatic-interpretation-of-otoliths-using-deep-learning

Recent advances in machine learning have brought forth methods that have been remarkably successful in a variety of settings, most notably in image analysis. These methods are now being applied to data analysis in marine sciences, where they have the potential to automate analysis that previously required manual curation. Here we adapt a machine learning model intended for object recognition to the task of estimating age from otolith images. The model is trained and validated on a collection of otolith images from Greenland halibut. We show that the precision of the model's age estimates is comparable to and may even surpass that of human experts. Age reading from otoliths is an important element in the management of many marine stocks, and automating this analysis is an important step to ensure consistency, lower cost, and increase scale.

Automatic-interpretation-of-otoliths-using-deep-learning

Training images used can be found: http://metadata.nmdc.no/metadata-api/landingpage/a3e59fb0e340a4f4a7ba76ea658b64b8