Yi Jin Toh ’25, Computer Science

Yi Jin Toh ’25

Major

Computer Science

Project Title

Machine Learning for Biodiversity Monitoring

Minor(s): Environmental Studies

Camera traps are crucial to biodiversity conservation, enabling data collection on wildlife populations without disrupting natural habitats or requiring intensive fieldwork. However, the vast amount of data often overwhelms manual processing, leading to overlooked conservation insights. While machine learning models are increasingly used to automate this analysis, they struggle with out-of-distribution (OOD) data — i.e., species or objects that were not seen during training — which can compromise prediction reliability. My internship focused on addressing this OOD challenge through machine learning research. I conducted a comparative study of state-of-the-art OOD detection methods using real-world camera trap data, evaluating their effectiveness in biodiversity monitoring. Additionally, I improved our team’s software and data engineering processes by developing image processing pipelines and revamping our database. This project was part of a broader initiative led by my supervisors at Conservation X Labs and their partners at Microsoft’s AI for Good Lab, who are leaders in the field working to establish best practices for applying machine learning on camera trap data. This experience has been deeply rewarding, as my work will directly impact real-world conservation efforts. The experience and connections I gained have prepared and further motivated me to continue pursuing conservation technology.