Can you  tell us more about your role and how you apply data science to your projects?

As a specialist in computer vision (CV) and artificial intelligence (AI), I support ecologists at the Institute for Biodiversity and Ecosystem Dynamics (IBED). My work involves designing and training deep learning models to tackle various ecological challenges. For example, I use GPS/IMU data from birds to classify different behaviors using deep learning techniques. Another project focuses on tracking fish in video footage. I also mentored students by creating tutorials and open-source notebooks to guide their learning.

Is there a project from the past year that you're particularly proud of?

It’s difficult to choose just one project because I’m proud of all of them! That said, supervising two PhD candidates who successfully published their research at ICCV, one of the top conferences in computer vision and AI, stands out. Their innovative work involved designing deep learning models for fish tracking and bird behavior analysis. Bird behavior studies can offer early warnings for environmental disasters, while fish tracking helps reveal the human impact on ecosystems. These projects illustrate the potential for AI to contribute to ecological research.

What do you like most about being a Data Science Centre (DSC) member?

The best part is the interdisciplinary connections. Being part of DSC has helped me connect with experts from various fields and recruit five master's students, whom I now supervise on data science projects related to ecology. These collaborations have been incredibly valuable for both my work and the students.

What is your favorite data science method?

As a CV and AI specialist, I primarily work with deep learning methods. I’m particularly drawn to cutting-edge algorithms like (editors addition: Neural Radiance Fields, a technique for 3D reconstruction) NeRF, generative models such as GANs (Generative Adversarial Networks) and VAEs (Variational Autoencoders), diffusion models, and transformer models. Each of these methods open up exciting possibilities for solving complex problems in both AI and ecology.

Are you more of a Python or R person, or do you prefer another language?

Definitely Python, as it's essential for deep learning. I haven’t worked much with R—I've only experimented with it briefly, and while it seems somewhat similar to Python, I find Python better suited for the tasks I work on.

Contact Fatemeh Karimi Nejadasl