Exploring the potential to use low cost imaging and an open source convolutional neural network detector to support stock assessment of the king scallop ()

Katja Ovchinnikova, Mark A. James, Tania Mendo, Matthew Dawkins, Jon Crall, Karen Boswarva

Abstract: Highlights•A relatively low cost camera system could be used to capture images of the scallop (Pecten maximus) for automated image recognition.•An annotated set of 3048 seabed images with the scallop P. maximus, was produced.•The VIAME tool kit scallop detectors were assessed for P. maximus.•A new detector based on the NatHarn algorithm and a Convolutional Neural Network was trained and applied to estimate P. maximus abundance.•A small dataset of scallop images obtained in a low-cost manner can be sufficient to train a reliable CNN model to detect P. maximus.