用少量数据和算力，让卫星图像自动识别树木类型与动态变化
Dargana: fine-tuning EarthPT for dynamic tree canopy mapping from space
- 微调地球观测基础模型，仅用3%数据和5%算力实现精准分类
- 10米分辨率下识别树冠覆盖，像素级准确率ROC-AUC达0.98
- 能捕捉篱笆、灌木等细小结构，适合长期森林监测与生态保护
我们提出Dargana，一种对EarthPT时序基础模型进行微调的变体，仅使用其预训练数据量的<3%和预训练计算量的5%，即可实现专业化。Dargana被微调用于生成10米分辨率的树冠覆盖定期分类，可区分针叶林与阔叶林。以英国康沃尔为测试案例，在未见卫星影像上，模型达到像素级ROC-AUC 0.98和PR-AUC 0.83。Dargana能识别低于训练样本限制的细小结构（如篱笆、砍伐区），并追踪新林地建立等时间变化。结果表明，像EarthPT这样的大型观测模型可通过轻量微调，实现从太空对土地覆盖进行精细、动态监测，为自然资本管理和保护提供可扩展工具。
原文摘要 · Abstract (English)
We present Dargana, a fine-tuned variant of the EarthPT time-series foundation model that achieves specialisation using <3% of its pre-training data volume and 5% of its pre-training compute. Dargana is fine-tuned to generate regularly updated classification of tree canopy cover at 10m resolution, distinguishing conifer and broadleaved tree types. Using Cornwall, UK, as a test case, the model achieves a pixel-level ROC-AUC of 0.98 and a PR-AUC of 0.83 on unseen satellite imagery. Dargana can identify fine structures like hedgerows and coppice below the training sample limit, and can track temporal changes to canopy cover such as new woodland establishment. Our results demonstrate how pre-trained Large Observation Models like EarthPT can be specialised for granular, dynamic land cover monitoring from space, providing a valuable, scalable tool for natural capital management and conservation.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。