Frognition: An AI solution for frog re-identification and species conservation efforts

The Problem

The conservation of the endangered Northern Leopard Frog was hindered by a manual re-identification process that required comparing new images against thousands of database records. This task was extremely time-consuming and relied on the availability of rare subject matter experts, creating a significant bottleneck in monitoring population health and trends efficiently.

The Solution

Frognition is an AI application that uses computer vision and deep learning to analyze frog photos and detect unique patterns, automatically re-identifying individual frogs without physical tags. The application was built using open source tools within existing budget.

The Impact

Frognition significantly reduces hours spent manually comparing images and helps track individual frogs over time to monitor populations and health without disturbing them.

Lessons Learned

Early approaches using self-supervised models weren't suitable, and data quality and quantity were major challenges. Shifting to a classification approach improved accuracy significantly.

Launch year: 2025