A Dual-Stream CLIP–ViT Framework for Open-Set Animal Re-Identification: Multi-Seed Ablation, Background-Bias Bracketing, and Query-Time Robustness Analysis
Journal of Imaging · 04.08.2026 · 12 (8) · 354
DOI: 10.3390/jimaging12080354 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Imaging, online erschienen am 04.08.2026. Verfasst von Ivan Melegatti Fernigrini und 1 weiteren Autor:innen. Beteiligte Einrichtung: School of Artificial Intelligence, Zhejiang University of Science and Technology (ZUST), Hangzhou 310023, China. Die Arbeit nennt 35 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Animal re-identification (Re-ID) asks whether two images show the same individual, a recognition task that fits naturally into applications such as reuniting lost pets with their owners. Existing methods report strong scores, but typically under a single seed, one mask granularity, and no query-time corruption analysis, leaving open whether the gains survive deployment. We propose a hierarchical framework decoupling localisation (a YOLOv8 soft-crop) from identity embedding: a dual-stream network fusing a frozen CLIP ViT-B/16 (learned projection) with a fine-tuned ViT-Base carrying L2-norm part attention, trained under ArcFace. On a combined cat+dog open-set benchmark of 173 identities, it attains Rank-1 0.9742/mAP 0.8597 over three seeds, surpassing a ViT-only ablation by +2.39 Rank-1 and +2.07 mAP. Open-set verification shows all configurations converge near 68% true acceptance at the strictest false-acceptance rate. A background-bias evaluation brackets the embedding’s background reliance between a bounding-box lower bound and a SAM-silhouette upper bound; a manual audit retains the reliable soft crop. A nine-corruption audit identifies down-sampling and motion blur as dominant. On PetFace, the architecture retrieves across 14,716 unseen identities and remains viable in a few-shot regime. A Descriptor Vector Exchange (DVE) extension is Pareto-dominated, traced to the ViT’s coarse feature map and architectural redundancy.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Artificial Intelligence, Zhejiang University of Science and Technology (ZUST), Hangzhou 310023, China.
ORCID-Profile: Ivan Melegatti Fernigrini, Bensheng Yun
Zitieren
APA
Melegatti Fernigrini, I., & Yun, B. (2026). A Dual-Stream CLIP–ViT Framework for Open-Set Animal Re-Identification: Multi-Seed Ablation, Background-Bias Bracketing, and Query-Time Robustness Analysis. Journal of Imaging, 12(8), 354. https://doi.org/10.3390/jimaging12080354
Vancouver
Melegatti Fernigrini I, Yun B. A Dual-Stream CLIP–ViT Framework for Open-Set Animal Re-Identification: Multi-Seed Ablation, Background-Bias Bracketing, and Query-Time Robustness Analysis. Journal of Imaging. 2026;12(8):354. doi:10.3390/jimaging12080354
BibTeX
@article{MelegattiFernigrini2026,
author = {Melegatti Fernigrini, Ivan and Yun, Bensheng},
title = {A Dual-Stream CLIP–ViT Framework for Open-Set Animal Re-Identification: Multi-Seed Ablation, Background-Bias Bracketing, and Query-Time Robustness Analysis},
journal = {Journal of Imaging},
year = {2026},
volume = {12},
number = {8},
pages = {354},
doi = {10.3390/jimaging12080354}
}
EUVIMED