Lab Blogs

Computer Vision and Conservation Science

Artificial intelligence, computer vision, machine learning, shape predictors, GPUs, face recognition….not usually terms you hear within a lab dedicated to conservation science! However, these technological advances are starting to become of interest to scientists outside the field of computer science. ‘Conservation technology’ is a rapidly emerging field which bridges the gap between hardware/software development, and the conservation of species and landscapes. The application of technical tools to address both new and pre-existing research questions has garnered the attention of the conservation science community. From drones and camera traps to automated methods of data processing, and the combination of the two, technological developments like this provide the opportunity for us to conduct research more efficiently, and maybe even effectively.

Here at the lab, we are working to combine deep learning approaches with ecological research techniques, to develop advanced methods to monitor grizzly bears. I work directly with software engineers (Ed Miller and Mary Nguyen, Hyperaptive, USA) who provide the technical expertise, while I am responsible for data collection, field trials, and eventually implementation. Together we form the BearID Project. We aim to develop face recognition technology for grizzly bears, driven by the current inability to recognize individuals from images alone. Individual ID of bears from images would open up a wealth of research opportunities for the species, worldwide. We also intend for this project to provide a benchmark for the development and application of facial recognition in other large mammals.

In recognition of our efforts so far, the BearID Project was recently awarded 2nd place in the WILDLABS #Tech4Wildlife photo challenge. This competition showcases projects that use conservation technology in the lab or field. This recognition was largely thanks to our collaborator Ed’s comprehensive blogging on each aspect of software development so far (read all about it here). This has given us fantastic online exposure for the project and further developed our connection with the WILDLABS.NET community.

By Melanie Clapham

Postdoctoral Fellow, ACS Lab