Research

Thesis

For my Ph.D. thesis, “Contour-Based Instance Recognition of Animals”, I developed CurvRank, a computer vision and machine learning algorithm that identifies individual African elephants, humpback whales and dolphins from photos of their ears, flukes, or fins. CurvRank uses the curvature of the fin as a unique signature to compute a ranking of likely matches from a database.

CurvRank has been integrated into WildBook, an ecological computer vision and data management software system developed by the non-profit Wild Me, and has been used by several ecological research groups to accelerate their matching efforts.

Publications

Computer vision venues, as first author

• Integral curvature representation and matching algorithms for identification of dolphins and whales
• Extracting identifying contours for African elephants and humpback whales using a learned appearance model

Other venues, as collaborating author

• Testing of two new automated fluke identification algorithms and comparison to non-automated methods for humpback whales
• Tendencias, vínculos migratorios y uso de hábitat de la ballena jorobada Megaptera novaeangliae (Borowski, 1781) en Guerrero, México

In the Media

• From binoculars to big data: Citizen scientists use emerging technology in the wild
• NVIDIA I am AI Docuseries, Episode 6: Running Wild for Nature Conservation - Wildbook
• How artificial intelligence is changing wildlife research