Multispecies Animal Re-ID Using a Large Community-Curated Dataset

TL;DR

This paper introduces a multi-species animal re-identification model trained on a large curated dataset, outperforming species-specific models and enabling effective zero-shot and fine-tuning for new species, with practical deployment in wildlife monitoring.

Contribution

The paper presents a single multi-species re-identification model trained on a large dataset, improving accuracy and flexibility over traditional species-specific models, and demonstrating strong zero-shot and fine-tuning performance.

Findings

12.5% average top-1 accuracy gain over species-specific models

19.2% top-1 improvement on unseen species compared to MegaDescriptor

Model deployed in large-scale wildlife monitoring system

Abstract

Recent work has established the ecological importance of developing algorithms for identifying animals individually from images. Typically, a separate algorithm is trained for each species, a natural step but one that creates significant barriers to wide-spread use: (1) each effort is expensive, requiring data collection, data curation, and model training, deployment, and maintenance, (2) there is little training data for many species, and (3) commonalities in appearance across species are not exploited. We propose an alternative approach focused on training multi-species individual identification (re-id) models. We construct a dataset that includes 49 species, 37K individual animals, and 225K images, using this data to train a single embedding network for all species. Our model employs an EfficientNetV2 backbone and a sub-center ArcFace loss function with dynamic margins. We evaluate the performance of this multispecies model in several ways. Most notably, we demonstrate that it consistently outperforms models trained separately on each species, achieving an average gain of 12.5% in top-1 accuracy. Furthermore, the model demonstrates strong zero-shot performance and fine-tuning capabilities for new species with limited training data, enabling effective curation of new species through both incremental addition of data to the training set and fine-tuning without the original data. Additionally, our model surpasses the recent MegaDescriptor on unseen species, averaging an 19.2% top-1 improvement per species and showing gains across all 33 species tested. The fully-featured code repository is publicly available on GitHub, and the feature extractor model can be accessed on HuggingFace for seamless integration with wildlife re-identification pipelines. The model is already in production use for 60+ species in a large-scale wildlife monitoring system.