Bearcam Companion is an experiment related to the BearID Project . BearID Project got its start many years ago when 2 engineers were watching the Explore.org bear cams and decided to try to build a machine learning application to identify bears. Those engineers got together with a conservation scientist in British Columbia. Together, we are working on methods to non-invasively identify brown bears in the wild using camera traps and the machine learning algorithms we are developing. You can read the back story here .
The team has always wanted to bring the project back to the bear cams. However, the machine learning models we have developed are very dependent on high quality photos, like those from a digital camera. The bear cam quality is getting better, but it's not quite good enough in general. The BearID model also identifies bears by their faces, which are not always visible on the cams. For these reasons, we have not been able to reliable apply the models to the bearcam feeds.
In March of 2022, one of those original engineers , Ed Miller , was accepted into the AWS Community Builders program (he's now an AWS Hero). As part of this experience, Ed has started developing a web application using Amazon Web Services and credits provided by the AWS. The application is called the Bearcam Companion, and has the following goals:
- Utilize AWS services and machine learning to enhance the bear viewing experience
- Help newer members of the bear cam community identify bears
- Enable experienced members of the community to share their knowledge of the bears
- Leverage community knowledge to teach the application to identify the bears
Try out the web application here: https://app.bearid.org/
We hope you find this project beneficial. Please help us continue to improve by identifying bears! For information on how to contribute, see the instructions page. Happy ID-ing!