发表机构
Xoople（Xoople）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文通过受控实验分析Tessera嵌入的时间敏感性，发现其性能与任务相关，时间覆盖可作为可调成本，能支撑近实时制图等应用。
AI 中文摘要
许多地球观测应用需要既精确又频繁更新的土地利用/土地覆盖图，但最强的地球观测基础模型从全年观测中构建其嵌入。我们对领先基础模型之一Tessera在土地利用/土地覆盖制图中的时间敏感性进行了受控研究。保持编码器冻结，我们在不同的观测窗口（从全年到单日）上重新计算其嵌入，将它们作为线性探测和UNet分割头的输入，在LUCAS、DynamicEarthNet和PASTIS-R数据集上将两者与从头训练的网络进行基准测试。我们表明，嵌入的价值取决于任务：在物候学区分类别的地方，如PASTIS-R的作物类型，它们达到58.3的平均交并比，比最佳从头训练模型高出约46%；在类别的时间稳定的地方（如DynamicEarthNet和LUCAS中的森林），基于嵌入的模型和从头训练的模型仅在全监督下表现相当，且在这两个数据集上，Tessera嵌入的标签效率明显更高。较短时间窗口下的性能下降是渐进的且取决于类别，将窗口从一年收缩到一个月，PASTIS-R的分割精度损失39%，而DynamicEarthNet仅损失5%；单日嵌入仍能以3.4倍的 chance level对LUCAS的土地覆盖进行分类。我们的研究表明，时间覆盖是可调整的成本而非固定前提，为近实时制图和更快的土地利用/土地覆盖更新周期等机制开辟了可能。
英文摘要
Many Earth Observation applications need land-use/land-cover maps that are both precise and frequently updated, yet the strongest Earth Observation foundation models build their embeddings from a full year of observations. We present a controlled study of the temporal sensitivity of Tessera, one of these leading foundation models, for land-use/land-cover mapping. Keeping the encoder frozen, we recompute its embeddings over varying observation windows, from a full year down to a single day. We use them as inputs to a linear probe and a UNet segmentation head, benchmarking both of them against from-scratch networks on LUCAS, DynamicEarthNet, and PASTIS-R datasets. We show that the value of the embeddings is task-dependent. Where classes are separated by phenology, as for the crop types of PASTIS-R, they reach a mean Intersection-over-Union of $58.3$, about $46\%$ above the best from-scratch model. Where classes are temporally stable (e.g., forests in DynamicEarthNet and LUCAS), embedding-based and from-scratch models match only under full supervision. On both datasets, Tessera embeddings remain markedly more label-efficient. Degradation under shorter temporal windows is gradual and class-dependent. Contracting the window from one year to one month costs $39\%$ of the segmentation accuracy on PASTIS-R but only $5\%$ on DynamicEarthNet. Single-day embeddings still classify land cover in LUCAS at $3.4$ times the chance level. Our study shows that temporal coverage is therefore a tunable cost rather than a fixed prerequisite, opening regimes such as near-real-time mapping and faster land-use/land-cover refresh cycles.
Comments15 pages
