研究人员开发了MetaPerch，这是一种新颖的生物声学基础模型，它利用了Xeno-Canto等公民科学平台的元数据。通过整合来自位置和时间数据的辅助监督信号，MetaPerch提高了物种识别性能，并增强了跨不同声学域的泛化能力。一项广泛的实证研究证明了九种不同元数据来源对十七个生物声学数据集的影响。 AI
影响 通过利用元数据增强生物声学中的物种识别能力，有望改进现实世界的被动声学监测。
排序理由 该集群包含一篇详细介绍新型生物声学基础模型的研究论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
完整方法见我们的编辑标准。
arXiv:2607.14072v1 Announce Type: new Abstract: Bioacoustic foundation models rely on large-scale citizen science platforms like Xeno-Canto for geographically and ecologically diverse data. Recent work has shown that supervision alone can produce SotA species detection models whe…
Bioacoustic foundation models rely on large-scale citizen science platforms like Xeno-Canto for geographically and ecologically diverse data. Recent work has shown that supervision alone can produce SotA species detection models when trained on this large-scale data -- however, t…
Bioacoustic foundation models rely on large-scale citizen science platforms like Xeno-Canto for geographically and ecologically diverse data. Recent work has shown that supervision alone can produce SotA species detection models when trained on this large-scale data -- however, t…