<b>Using Computer Vision to Age Otoliths of Antarctic Toothfish</b>

<p dir="ltr">Antarctic toothfish (<i>Dissostichus mawsoni</i>) are a keystone species in the Southern Ocean ecosystem at risk of overfishing due to their high commercial demand. Understanding the Antarctic toothfish’s age distribution is crucial for enacting regulatory policies to protect the species, and is traditionally done manually by aging their otoliths. Otoliths are calcified structures in the fish’s ear that contain growth zones similar to tree rings, which are counted to determine the age of a fish in years. The <b>Otolith-Ages</b> dataset was constructed from 1302 images of otoliths, and was used to train aging models with regression and classification formulations. Resulting aging models were capable of aging otoliths with an average percent error of 6.3%, coming within 1.7% of trained human readers. Another dataset, <b>Otolith-</b><b>Edges</b>, was constructed from a subset of 58 otolith images annotated with edge maps of their growth zones to facilitate future work on aging otoliths using edge detection methods.</p>