Artificial Intelligence can help to identify southern right whale individuals... and you can help us do it!
Southern right whales are characterised by the white patches (callosities) on their head. These callosities are inhabited with tiny crustaceans called cyamids or whale lice, which gives the callosity a typical whitish colouration.
The pattern these callosities form on the whale’s head is unique for each individual, and is used by researchers to identify each whale individually.
In some of the wintering areas, long-term photo-identification studies have been ongoing since the 1970s. While methodologies for these long-term studies vary, most involve annual aerial surveys that count and photo-identify southern right whales in coastal waters based on overhead images of the callosity pattern on their head. These studies have provided a wealth of information on the behaviour, population abundance and trends over the past five decades.
Due to the large sizes, simple manual identification through visual comparison of images is not feasible, and semi-automated processes were implemented in the early 2000s. However, these systems require manual annotation and photo manipulation, making the process of photo-identification very time-consuming and labour intensive.
With the widespread use of digital photography and drones for whale research, tens of thousands of right whale photographs are taken each year. In fact, over 13,500 individuals have been identified by all consortium partners combined.
The inability to process such large volumes of photo-identification data has led to bottlenecks in analyses of population trends, movement patterns, population connectivity and a global population assessment
Therefore, we are working collaboratively to leverage artificial intelligence (AI) to develop a fully automated photo-identification system especially designed for southern right whales.
However, to fully develop and implement this technology, there is a need for financial resources to support the work of AI developers.
If you want to be a project partner, please don’t hesitate to get in touch (els.vermeulen@up.ac.za)
Project partners
We are very grateful to the following sponsors of this project:
