Thread by @bearid_project

We use #Tech4Wildlife for our project developing automated methods of individual ID for brown bears and other wildlife. We use a combination of #cameratrapping, visual observations, and #machinelearning with collaborative partnerships to make this happen. @WILDLABSNET 1/

Despite the pandemic, 2020 has been a big year for our project. Our wonderful partners @knightinlet @wildbearlodge @nanwakolas took a central role in getting #cameratraps up and running for the field season. #Tech4Wildlife @WILDLABSNET 2/

We conducted some testing of our BearID software in real time using @Docker and the help of some #grizzlybears. We are using a population of bears that have been studied for >20yrs to assess the robustness of our #DeepLearning models. #Tech4Wildlife @WILDLABSNET 3/

We recorded known individuals and then ran images through the #BearID pipeline in @Docker to both detect the bear's face and identify the individual. Luckily we had some dry days and fairly calm seas for this part! #Tech4Wildlife @WILDLABSNET 4/

Then it was back to our field base, to look at the results further. We were looking to see how variables such as lighting and facial pose affect ID accuracy. This work continues. #Tech4Wildlife @WILDLABSNET 5/

We wrapped up 2020 with the publication of our full BearID pipeline (v 20.05) in the peer-reviewed journal, @Ecol_Evol. We put out a press release and received some fantastic coverage of our project - we even made it into the @nytimes! #Tech4Wildlife @WILDLABSNET 6/

Moving forward into 2021 we are focusing on applying BearID to #cameratrap images, as well as other species where individual photo-ID can be difficult. #Tech4Wildlife @WILDLABSNET 7/7