Koala Detection in Wildfire

The aim of this project is to detect the animals living in the wild and identify the impact of bushfires on their health. With the recent event of bushfires happened across Australia, not only it affected the human lives but also impacted the animals living the forest. According to World Wide Fund for Nature or WWF, nearly 3 billion animals killed or harmed with the Australian bushfire in 2019-20 (WWF 2020). This is a grave number and can sweep some very endangered species from our planet. Thus, to reduce these issues and provide wildlife support, our industry partner and we planned to use Artificial Intelligence detect all the injured animals or any species who need immediate medical assistance. We will be combining the power of Deep Learning and Drone flight to capture the live footage of animals.

The goal of the project is to evaluate various object detection models that can be used for identifying various classes like animals in thermal images captured through the drone. The application of this project would help identify animals accurately through the drone footage so that we can provide medical attention/rehabilitation to them as per the requirements.

We decided to use thermal images for our problem as the animals are most active during the night time, and it is hard to capture precise details using a standard camera in dark conditions. Therefore, by capturing the heat signatures of animals, we would be able to better train our deep learning model for identifying animals at night when compared to the model trained using the normal images irrespective of the lighting conditions.