Jess Tam

I completed my BSc and BSc (Hons) in Biology here at UNSW. My honours thesis was supervised by Shinichi Nakagawa and Will Cornwell, where I investigated the bibliometric biases in mammalian research in the scientific literature. My research interests include conservation technologies, real-world applications of computer vision, and bibliometric analyses. When I'm not working, I'm probably enjoying some delicious food.

Project: Applications of computer vision in camera trap data processing

Supervised by: Richard Kingsford, Arcot Sowmya, Shinichi Nakagawa (University of Alberta), Hideo Saito (Keio University)

Project Description: Computer vision (CV) is an area of research that studies how machines process (or 'see') our visual world. While many fields have now adopted CV tools for automation, such as in manufactoring and medicine, ecologists have been slow to apply these tools. In addition, existing models struggle to generalise to camera trap data as the subjects and backgrounds of the imagery are often quite different from the data that they were trained on, which is usually similar to those from the ImageNet or COCO datasets.

My research focuses on using existing CV tools to automatically process motion-triggered camera trap images in order for ecologists to make inferences, such as estimating population densities of wildlife. My main collaborators are NSW National Parks and Wildlife Service and Wild Deserts, who have kindly provided their labelled datasets. We've come across many different species of terrestrial wildlife in our datasets, but mammals are my main study species for building and testing models.

You can find out more about my research on my website.