提出LEPA模型，精准预测卫星图像几何变换后的特征表示。
LEPA: Learning Geometric Equivariance in Satellite Remote Sensing Data with a Predictive Architecture
- 用几何变换条件预测替代平均插值，学习特征的几何等变性。
- 在HLS数据上将匹配精度（MRR）从0.2提升至0.8以上。
- 适合需要灵活区域分析的遥感应用，避免重新编码计算。
地理空间基础模型提供预计算的嵌入向量，作为大规模卫星遥感数据的紧凑特征表示。尽管这些嵌入可缓解数据传输瓶颈和计算成本，但用户定义的兴趣区域与固定预计算嵌入网格之间仍存在几何不匹配问题。标准隐空间插值在此场景下不可靠，因嵌入流形高度非凸，导致生成的表示不符合真实输入。我们通过Prithvi-EO-2.0验证了插值方法的局限性。为此，提出一种学习等变性预测架构（LEPA）。LEPA不采用向量平均，而是以几何增强为条件，直接预测变换后的嵌入。我们在NASA/USGS Harmonized Landsat-Sentinel（HLS）影像和ImageNet-1k上评估该方法。实验表明，标准插值的均倒数排名（MRR）低于0.2，而LEPA将MRR提升至0.8以上，实现无需重新编码的精确几何调整。
原文摘要 · Abstract (English)
Geospatial foundation models provide precomputed embeddings that serve as compact feature vectors for large-scale satellite remote sensing data. While these embeddings can reduce data-transfer bottlenecks and computational costs, Earth observation (EO) applications can still face geometric mismatches between user-defined areas of interest and the fixed precomputed embedding grid. Standard latent-space interpolation is unreliable in this setting because the embedding manifold is highly non-convex, yielding representations that do not correspond to realistic inputs. We verify this using Prithvi-EO-2.0 to understand the shortcomings of interpolation applied to patch embeddings. As a substitute, we propose a Learned Equivariance-Predicting Architecture (LEPA). Instead of averaging vectors, LEPA conditions a predictor on geometric augmentations to directly predict the transformed embedding. We evaluate LEPA on NASA/USGS Harmonized Landsat-Sentinel (HLS) imagery and ImageNet-1k. Experiments show that standard interpolation achieves a mean reciprocal rank (MRR) below 0.2, whereas LEPA increases MRR to over 0.8, enabling accurate geometric adjustment without re-encoding.
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。