Tracking individual bears through time

In 2018 the ACS Lab was awarded a 2-year NSERC Collaborative Research and Development grant to support our work developing automated facial recognition for brown bears. Our aim is for this technology to be used to improve non-invasive monitoring techniques for bears and other large mammals. Part of this work involves a lengthy amount of time in the field.

We have two main field sites for this project, both are supported by eco-tourism partners. Our first site is in Knight Inlet, on the south-central coast of BC. I have been working at this site for ten years alongside Knight Inlet Lodge, who co-fund and help to facilitate the research. We are continually grateful to the Da’naxda’xw/Awaetlala First Nation for allowing us to study bears within their traditional territory.

Our second site is in the interior of BC, within the Selkirk-Purcell mountain ranges. Our work in this area is co-funded and supported by Grizzly Bear Ranch.

Our 2018 field season ran from April-October. The primary aim of this season was to collect training and test images of bears for our facial recognition system. At our Knight Inlet field site, a catalogue of different individual bears has been maintained since research began in the late 1990s. Much of this prior work involved behavioural studies, which closely monitored ‘resident’ and returning individuals to the area. This work provided us with a baseline database of images of individuals, which we are currently building upon.

One benefit to us of long-term documentation of individuals is that we can use these images to test how our recognition system performs when analyzing images of the same bears over time. Bears can change dramatically in their physical appearance between spring and late-fall and also over successive years, particularly for younger bears. Therefore, tracking and documenting these changes over time and using these images in training will help to ensure that our facial recognition system performs well on wild bear populations.