arXiv 2606.213092026-06-23cs.CV

WildBox: 非洲稀树草原野生动物航拍单目3D检测数据集与基准

WildBox: A Dataset and Benchmark for Aerial Monocular 3D Detection of African Savanna Wildlife

D Optical Metrology Unit, Fondazione Bruno Kessler（布鲁诺·凯斯勒基金会3D光学计量部）；Computer Vision and Machine Learning Systems group, Institute for Geoinformatics, University of Muenster（明斯特大学地理信息学研究所计算机视觉与机器学习系统组）；School of Civil, Aerospace and Design Engineering, University of Bristol（布里斯托大学土木、航空航天与设计工程学院）；Department of Biology, University of Southern Denmark（南丹麦大学生物学系）；The Ohio State University（俄亥俄州立大学）；Department of Collective Behavior, Max Planck Institute of Animal Behavior（马克斯·普朗克动物行为研究所集体行为系）；University of Konstanz（康斯坦茨大学）；Centre for the Advanced Study of Collective Behaviour, University of Konstanz（康斯坦茨大学集体行为高级研究中心）；Department of Biology, University of Konstanz（康斯坦茨大学生物学系）；Environmental Computational Science and Earth Observation Laboratory, EPFL（瑞士联邦理工学院环境计算科学与地球观测实验室）

Vandita Shukla, Kilian Meier, Lucie Laporte-Devylder, Camille Rondeau Saint-Jean, Jenna M. Kline, Blair R. Costelloe, Devis Tuia, Fabio Remondino, Benjamin Risse

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出WildBox数据集，含23.7万3D框标注，评估两种开放词汇单目3D架构，发现零样本3D检测失败，微调后恢复性能，深度估计是主要瓶颈。

摘要：

我们介绍了WildBox，一个用于从无人机视频中进行野生动物单目3D检测的数据集和基准，包含237,505个3D边界框标注，涵盖七种非洲稀树草原物种，分为六个基准类别。标注采用KITTI/Omni3D兼容格式，位于每个片段尺度归一化的相机坐标系中，并在每个片段内保持实例身份。我们在零样本、真实2D框提示和监督微调协议下评估了两种开放词汇单目3D架构：OVMono3D-LIFT和DetAny3D。开放词汇2D基础模型提供了可用的零样本野生动物定位（50.55 AP@50），但零样本3D检测在所有测试的2D输入条件下（包括真实2D框提示）均崩溃为0.00 AP，从而将失败隔离到3D阶段。在WildBox上微调将性能恢复到8.68 +/- 0.47 AP-BEV@0.50和13.17 +/- 0.69 AP3D宏平均。微调后深度贡献了归一化Hausdorff距离的84%，在零样本中超过99%，将单目航拍深度识别为该场景下的主要开放问题。一种由粗到精的课程学习（即在合并的斑马类别上预训练，然后在细纹斑马/平原斑马分类上微调）以更少的总计算量提高了宏平均3D性能，在两个斑马子类上增益最大。WildBox随视频级划分、评估代码和基线检查点一起发布，以推动无人机视频中3D野生动物感知的进展。

英文摘要：

We introduce WildBox, a dataset and benchmark for monocular 3D detection of wildlife from drone video, comprising 237,505 3D bounding box annotations across seven African savanna species grouped into six benchmark classes. Annotations follow a KITTI/Omni3D-compatible format in a per-segment scale-normalised camera frame, with instance identities maintained across each segment. We evaluate two open-vocabulary monocular 3D architectures, OVMono3D-LIFT and DetAny3D, under zero-shot, ground-truth 2D box prompt, and supervised fine-tuning protocols. Open-vocabulary 2D foundation models provide usable zero-shot wildlife localisation (50.55 AP@50), but zero-shot 3D detection collapses to 0.00 AP across both architectures and every 2D-input condition tested, including ground-truth 2D box prompts, thus isolating the failure to the 3D stage. Fine-tuning on WildBox recovers performance to 8.68 +/- 0.47 AP-BEV@0.50 and 13.17 +/- 0.69 AP3D macro. Depth contributes 84% of normalised Hausdorff distance after fine-tuning and over 99% in zero-shot, identifying monocular aerial depth as the dominant open problem in this regime. A coarse-to-fine curriculum, i.e. pretraining on a merged zebra class before fine-tuning on the Grevy's/plains split, improves macro 3D performance with less total compute, with the largest gains on the two zebra subclasses. WildBox is released with video-level splits, evaluation code, and baseline checkpoints to enable progress in 3D wildlife perception from drone video.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。