发表机构
Kansai University; Wakayama Fruit Tree Experiment Station; Mie University; Fukushima University; The University of Tokyo（关西大学; 和歌山果树试验站; 三重大学; 福岛大学; 东京大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对日本里山夜间红外相机陷阱数据，提出多轴基准SatoyamaCT，联合评估域泛化、新物种检测与选择性预测，并验证不确定性分诊可显著恢复区域性能。
AI 中文摘要
日本里山地区农田与森林交错，梅花鹿、野猪和日本猕猴造成的农作物及森林损害是一个严重的经济问题。相机陷阱可实现可扩展的监测，然而现有基准在分布内评估识别性能，很少优先考虑夜间红外图像，且未在单一数据集上联合解决区域域偏移、新物种检测和基于不确定性的分诊问题。我们引入了里山相机陷阱数据集（SatoyamaCT），包含来自三个里山区域的以夜间红外为主的相机陷阱的12,642个经专家验证的农作物图像。一种结合BioCLIP嵌入与置信度排序确认的迭代标注协议减少了专家工作量，同时所有标签均由领域生态学家验证；物种级标注者间一致性达到Cohen's kappa = 0.900。一个多轴协议在基于捕获事件的泄漏控制下，联合评估跨区域、相机位置和光照的域泛化、开放集新物种检测以及选择性预测。难度区分为不同的失败模式：在Wakayama地区，四个骨干网络及包括CORAL、DANN和GroupDRO在内的域泛化方法中，数据固有的区域差距持续存在16至27个百分点，开放集检测总体达到AUROC 0.93至0.96，但在Wakayama地区与分类性能同时下降。选择性预测在50%覆盖率下将Wakayama地区准确率从0.593恢复至0.815，支持面向农林复合生态系统实际害虫监测的不确定性感知分诊工作流程。
英文摘要
Crop and forest damage from sika deer, wild boar, and Japanese macaque is a serious economic problem in Japanese satoyama, where farmland and forest intermingle. Camera traps enable scalable monitoring, yet existing benchmarks evaluate recognition in-distribution, rarely prioritize night infrared imagery, and do not jointly address regional domain shift, novel-species detection, and uncertainty-based triage on a single dataset. We introduce the Satoyama Camera Trap Dataset (SatoyamaCT), 12,642 expert-verified crops from night-IR-dominant camera traps across three satoyama regions. An iterative annotation protocol combining BioCLIP embeddings with confidence-ordered confirmation reduced expert effort while all labels were verified by domain ecologists; species-level inter-annotator agreement reached Cohen's kappa = 0.900. A multi-axis protocol jointly evaluates domain generalization across region, camera placement, and illumination; open-set novel-species detection; and selective prediction, all under capture-event-based leakage control. Difficulty separates into distinct failure modes: a data-inherent regional gap of 16 to 27 percentage points persists in Wakayama across four backbones and domain-generalization methods including CORAL, DANN, and GroupDRO, and open-set detection reaches AUROC 0.93 to 0.96 overall yet degrades jointly with classification in Wakayama. Selective prediction recovers Wakayama accuracy from 0.593 to 0.815 at 50% coverage, supporting an uncertainty-aware triage workflow for practical pest monitoring in agroforestry.
