DS@GT AnimalCLEF: Triplet Learning over ViT Manifolds with Nearest Neighbor Classification for Animal Re-identification

2025-09-15

AI Summary

This study addresses the challenge of improving feature embedding quality and cross-domain adaptability in animal individual re-identification, particularly under few-shot settings. We propose a metric learning framework built upon domain-specific pretrained backbones—DINOv2 and MegaDescriptor—incorporating a triplet-loss-based projection head, a k-nearest neighbors classifier, and a robust thresholding mechanism to distinguish known from novel individuals. Our key finding is that generic visual representations are ill-suited for fine-grained animal re-ID, whereas domain-specific pretraining substantially enhances few-shot performance; moreover, post-hoc metric learning gains are fundamentally constrained by backbone characteristics. Experiments on AnimalCLEF 2025 show that triplet learning boosts MegaDescriptor’s performance by +0.13 in mean average precision, yet yields only +0.03 for DINOv2—demonstrating that backbone domain adaptability constitutes both the primary performance bottleneck and a critical axis for innovation.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Representation Learning for Vision
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

This paper details the DS@GT team's entry for the AnimalCLEF 2025 re-identification challenge. Our key finding is that the effectiveness of post-hoc metric learning is highly contingent on the initial quality and domain-specificity of the backbone embeddings. We compare a general-purpose model (DINOv2) with a domain-specific model (MegaDescriptor) as a backbone. A K-Nearest Neighbor classifier with robust thresholding then identifies known individuals or flags new ones. While a triplet-learning projection head improved the performance of the specialized MegaDescriptor model by 0.13 points, it yielded minimal gains (0.03) for the general-purpose DINOv2 on averaged BAKS and BAUS. We demonstrate that the general-purpose manifold is more difficult to reshape for fine-grained tasks, as evidenced by stagnant validation loss and qualitative visualizations. This work highlights the critical limitations of refining general-purpose features for specialized, limited-data re-ID tasks and underscores the importance of domain-specific pre-training. The implementation for this work is publicly available at github.com/dsgt-arc/animalclef-2025.

Problem

• Compares general-purpose and domain-specific models for animal re-identification
• Evaluates triplet learning effectiveness on different backbone embedding manifolds
• Addresses fine-grained recognition challenges in limited-data re-ID tasks

Innovation

• Domain-specific MegaDescriptor backbone for embeddings
• Triplet-learning projection head improving performance
• K-Nearest Neighbor classifier with thresholding