Leatherbacks Matching by Automated Image Recognition

Eric J. Pauwels; Paul M. Zeeuw; Danielle M. Bounantony

doi:10.1007/978-3-540-70720-2_32

Leatherbacks Matching by Automated Image Recognition

Eric J. Pauwels, Paul M. de Zeeuw, Danielle M. Bounantony

Source

Lecture Notes in Computer Science > Advances in Data Mining. Medical Applications, E-Commerce, Marketing, and Theoretical Aspects > Aspects of Data Mining > 417-425

Abstract

We describe a method that performs automated recognition of individual laetherback turtles within a large nesting population. With only minimal preprocessing required of the user, we prove able to produce unsupervised matching results. The matching is based on the Scale-Invariant Feature Transform by Lowe. A strict condition posed by biologists reads that matches should not be missed (no false negatives). A robust criterion is defined to meet this requirement. Results are reported for a considerable sample of leatherbacks.