Our tutorial draws on a diverse and vibrant community of research, tools, and practical insights. Whether you’re just getting started or already an experienced practitioner, you’ll find curated talks, seminal publications, and open-source toolkits designed to help you dive in—and discover something new.
❗Feel free to ping us if something is missing ❗
For a general introduction to animal Re-ID, see the following papers:
Schneider, S., Taylor, G. W., Linquist, S., & Kremer, S. C. (2019). Past, present and future approaches using computer vision for animal re‐identification from camera trap data. Methods in Ecology and Evolution, 10(4), 461-470.
Tuia, D., Kellenberger, B., Beery, S., Costelloe, B. R., Zuffi, S., Risse, B., ... & Berger-Wolf, T. (2022). Perspectives in machine learning for wildlife conservation. Nature communications, 13(1), 792.
For an introduction to animal Re-ID methods:
Ravoor, P. C., & Sudarshan, T. S. B. (2020). Deep learning methods for multi-species animal re-identification and tracking–a survey. Computer Science Review, 38, 100289.
For an introduction to animal Re-ID datasets:
Čermák, V., Picek, L., Adam, L., & Papafitsoros, K. (2024). WildlifeDatasets: An open-source toolkit for animal re-identification. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (pp. 5953-5963).
Papers on interpreting and explaining Re-ID models:
Shrack, L., Haucke, T., Salaün, A., Subramonian, A., & Beery, S. (2025). Pairwise Matching of Intermediate Representations for Fine-grained Explainability. arXiv.
For an introduction to human-in-the-loop animal Re-ID systems:
Kulits, P., Wall, J., Bedetti, A., Henley, M., & Beery, S. (2021). ElephantBook: A Semi-Automated Human-in-the-Loop System for Elephant Re-Identification. ACM SIGCAS Conference on Computing and Sustainable Societies (COMPASS).
Perez, G., Sheldon, D., Van Horn, G., & Maji, S. (2024, September). Human-in-the-Loop Visual Re-ID for Population Size Estimation. European Conference on Computer Vision (pp. 185-202).
Tanya Berger-Wolf
Sara Beery
Vojtěch Čermák
Jason Holmberg
Risa Shinoda
Ekaterina Nepovinnykh
Several models were developed as general-purpose models to recognise multiple animal species.
MegaDescriptor builds on the public datasets (see WildlifeDatasets below) and creates a general-purpose tool for animal ReID.
MiewID is an alternative to MegaDescriptor trained by using Wild Me private data collections.
Plugins for WildBook provide implementations of multiple methods including Pose Invariant Embeddings (whales, dolphins, sharks, manta rays), finFindR (whales, dolphins, sharks), MantaMatcher (giant manta rays), and HotSpotter (giraffes, zebras, leopards, cheetahs, jaguars, lynx, wild cats, hyenas, wild dogs, whales, dolphins, seals, turtles, salamanders, toads, manta rays, groupers, seadragons).
The software development is only possible with the abundance of high-quality animal ReID datasets.
WildlifeDatasets allows straightforward access to 44 publicly available single-species animal ReID datasets and provides a wide variety of methods for dataset pre-processing, performance analysis, and model fine-tuning.
HappyWhale is the largest Re-ID database with over 1M images of over 123k individual whales. Part of the database was released publicly as a Kaggle competition and attracted huge attention with over 10k competition entrants.
There are several multi-species datasets.
WildlifeReID-10k combines 37 datasets from WildlifeDatasets with permissible licences and creates a large-scale dataset with 10k individuals of mostly wild animals such as marine turtles, primates, birds, African herbivores, marine mammals, and domestic animals. It acts as a benchmark for both closed-set and open-set animal ReID.
PetFace is a large-scale animal face re-identification dataset of more than 257k mostly domestic individuals across 13 families, including cats, dogs, hamsters, hedgehogs, and rabbits.
Citizen science enables large-scale annotation of animal images, which is essential for training and validating re-identification models. It frees researchers' time and fosters public engagement in conservation by turning volunteers into active contributors to wildlife monitoring efforts.
WildBook is an open-source, AI-powered platform developed by Wild Me that enables wildlife researchers and citizen scientists to identify and track individual animals using photographs. Users upload photos of wildlife sightings, and the system matches individuals against a global database using unique patterns (e.g., stripes, spots). Wildbook supports numerous species and projects, like whale sharks, giraffes, zebras, and cheetahs, and facilitates non-invasive population monitoring for conservation purposes.
Zooniverse is the world’s largest platform for people-powered research, enabling volunteers to assist with scientific projects across disciplines, including wildlife conservation. In ecology projects, users classify species, behaviours or individual identification. The Zooniverse project Snapshots at Sea - Whale Identification acts as a citizen science branch of the HappyWhale project. Volunteers assist by marking distinguishing features on these images, contributing to a database that helps track whale migrations and population dynamics.
Additional citizen science databases, such as GBIF or iNaturalist, are of great use for the conservation of nature, but are not primarily interested in animal ReID.