Using Machine Learning For Automatic Fish Species Identification

As fisheries data becomes easier to obtain from anglers through mobile phone applications, such as MyCatch, it presents both opportunities and challenges to researchers who want to use these data to conduct their own research projects. A major challenge is being able to organize, classify and process this often overwhelming stream of data. Machine learning methods present opportunities to automate the process and help researchers answer important questions about our fisheries.

In this study, MyCatch collaborated with Asta Audzijonyte, Catarina Silva, Justas Dainys and Vincentas Vienozinskis to develop a “Scalable Open-Source Framework for Machine Learning-Based Image Collection, Annotation and Classification: A Case Study for Automatic FishSpecies Identification“. The paper has recently been published in the journal Sustainability, and is publicly available for download here.

_{Header image by Angler’s Atlas member DeadDriftMedia}