arXiv cs.AI· William Grolleau, Achraf Chaouch, Astrid Sabourin, Guillaume Lapouge, Catherine Achard·· 4 小时前AI 评分36
MOO：面向牛只重识别视角分析的多视角观测数据集
MOO: A Multi-view Oriented Observations Dataset for Viewpoint Analysis in Cattle Re-Identification
AI 导读
研究者发布多视角观测数据集 MOO，用于空中-地面牛只重识别（AG-ReID）的视角变化分析。数据集包含 1,000 头牛个体、128 个均匀采样视角，共 128,000 张标注图像，为合成数据。基于该数据集，研究量化了俯仰角影响并找到关键高度阈值，超过该阈值模型对未见视角的泛化显著更好；在零样本和监督设置下验证了向 4 个真实牛只数据集的可迁移性。MOO 已公开发布。
正文 · 原文
Abstract:Animal re-identification (ReID) faces critical challenges due to viewpoint variations, particularly in Aerial-Ground (AG-ReID) settings where models must match individuals across drastic elevation changes. However, existing datasets lack the precise angular annotations required to systematically analyze these geometric variations. To address this, we introduce the Multi-view Oriented Observation (MOO) dataset, a large-scale synthetic AG-ReID dataset of $1,000$ cattle individuals captured from $128$ uniformly sampled viewpoints ($128,000$ annotated images). Using this controlled dataset, we quantify the influence of elevation and identify a critical elevation threshold, above which models generalize significantly better to unseen views. Finally, we validate the transferability to real-world applications in both zero-shot and supervised settings, demonstrating performance gains across four real-world cattle datasets and confirming that synthetic geometric priors effectively bridge the domain gap. Collectively, this dataset and analysis lay the foundation for future model development in cross-view animal ReID. MOO is publicly available at this https URL.
| Comments: | 31 pages, 10 figures, extended version of the paper accepted to the CVPR 2026 Workshop on Computer Vision for Animal Behavior Tracking and Modeling (CV4Animals) |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2603.04314 [cs.CV] |
| (or arXiv:2603.04314v3 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2603.04314 arXiv-issued DOI via DataCite |
Submission history
From: William Grolleau [view email]
[v1]
Wed, 4 Mar 2026 17:36:26 UTC (5,160 KB)
[v2]
Thu, 28 May 2026 15:13:29 UTC (5,161 KB)
[v3]
Thu, 5 Nov 2026 10:14:23 UTC (4,853 KB)
来源：arXiv cs.AI · arxiv.org