Related Experiment Video
Updated: Jun 1, 2026
09:47
Spotting Cheetahs: Identifying Individuals by Their Footprints
Published on: May 1, 2016
An accurate, efficient, and accessible AI-powered solution for wildlife re-identification in conservation.
Shahrzad Gholami1, Derek E Lee2, Caleb Robinson3
1AI for Good Research Lab, Microsoft, One Microsoft Way, Building 124, Redmond, WA, 98052, USA. sgholami@microsoft.com.
Scientific Reports
|May 30, 2026
Summary
A new AI system, GIRAFFE, automates wildlife re-identification for ecological studies. This tool significantly reduces manual effort and enhances accuracy in tracking animal populations and informing conservation strategies.
Area of Science:
- Ecology
- Computer Science
- Artificial Intelligence
Background:
- Accurate wildlife re-identification is crucial for ecological research, including population estimation and behavioral studies.
- Existing methods often require extensive manual labeling, limiting scalability and efficiency.
- Automated systems are needed to streamline data collection and analysis for conservation efforts.
Purpose of the Study:
- To introduce GIRAFFE (Generalized Image-based Re-identification using AI for Fauna Feature Extraction), an automated system for wildlife re-identification.
- To enable efficient and accurate identification of individual animals, starting with giraffes, for ecological research.
- To reduce manual effort in wildlife data analysis and improve the cost-effectiveness of conservation studies.
Main Methods:
- GIRAFFE utilizes local feature matching for identifying known individuals and partitioning unknown ones for large-scale annotation.
- A user interface is provided for dataset curation and analysis of repeat survey data by both technical and non-technical users.
- The system automates key steps in the re-identification pipeline, contrasting with traditional manual labeling approaches.
Main Results:
- The GIRAFFE system achieved over 0.9 accuracy across nine standard metrics on real-world giraffe datasets.
- The system demonstrated significant speed improvements, running 120 times faster than baseline methods.
- A 132-fold improvement in cost-effectiveness was observed, alongside maintained accuracy and interpretability.
Conclusions:
- GIRAFFE offers an automated, accurate, and cost-effective solution for wildlife re-identification, extensible to various species.
- The system supports endangered species tracking, population dynamics analysis, and data-driven conservation strategies.
- Automation of re-identification processes significantly reduces manual workload while enhancing research capabilities.
