21nov1:00 pm2:00 pmAssessing the performance of open source, semi-automated pattern recognition software for harbour seal (Phoca vitulina) photo-identification dataIzzy Langley1:00 pm - 2:00 pm SOI Lecture Theatre Speaker organisation: University of St Andrews
Photo-identification (photo-ID) is a well-established, non-invasive, and cost-effective technique to collect longitudinal data of species that can be individually recognized based on natural markings. This field has been revolutionised by
Photo-identification (photo-ID) is a well-established, non-invasive, and cost-effective technique to collect longitudinal data of species that can be individually recognized based on natural markings. This field has been revolutionised by computer-aided software for pattern recognition, speeding up the processing of large numbers of images. Freely available algorithms exist for a wide range of species, but the choice of algorithm can have significant effects on the accuracy of the individual histories and therefore derived demographic parameter estimates. We tested the performance of three open source, semi-automated pattern recognition algorithms on harbour seal photo-ID data: ExtractCompare, Wild.ID and I3S Pattern. We also tested for the effects of database size, data collection method (camera-lens versus camera-scope) and data type (aspect of the body) to inform future fieldwork protocols. All algorithms generate similarity scores between pairs of images and the top 20 potential matches were inspected. The false rejection rate (FRR) of each software and the cumulative density function (CDF) for the ranks of true matches were calculated. Software algorithms have their own specific limitations, but ExtractCompare is currently the only pattern recognition software that accounts for orientation and torsion by using a 3-dimensional model. However, by selecting an area of the body where torsion is limited (i.e. the head), preliminary analysis suggests Wild.ID outperforms both ExtractCompare and I3S Pattern in harbour seal photo–ID data processing. As the analysis is still ongoing, the authors are hopeful to receive feedback on how the results of this study should be modelled to provide the most useful and best value outputs.
(Thursday) 1:00 pm - 2:00 pm
SOI Lecture Theatre
Gatty Marine Laboratory, St Andrews
University of St Andrews