卫星预训练并非万能：通用与地球观测基础模型的配对对比研究
Satellite Pretraining Is Not One Thing: A Matched-Pair Study of Generic and Earth-Observation Foundation Models for Scene Classification
AI 导读
一项配对对比研究考察地球观测（EO）预训练模型是否优于通用基础模型，在 14 个冻结编码器、5 个 RGB 场景分类数据集上评估 5 组配对（如 DINO/SatDINO、MAE/SatMAE、CLIP/RemoteCLIP）。
正文 · 原文
Abstract:Geospatial foundation models are pretrained on satellite and aerial imagery, but whether such an encoder beats a strong generic one for few-label land-use/land-cover mapping remains unclear, as prior comparisons rarely isolate the pretraining data. We study five matched pairs of Earth-observation (EO) and generic encoders sharing objective and architecture (DINO/SatDINO, DINOv3/DINOv3-SAT, MAE/SatMAE, MAE/Scale-MAE, CLIP/RemoteCLIP) within a benchmark of 14 frozen encoders on five RGB scene-classification datasets (0.06-30 m GSD). With native preprocessing, we evaluate k-NN, linear and attentive probes and image retrieval over six low-label splits, using paired per-image significance tests. EO pretraining is not uniformly helpful: on most datasets and metrics, it significantly improves three pairs (SatDINO, Scale-MAE, RemoteCLIP, the latter evaluated only on unseen datasets) and significantly hurts two (linear probe: DINOv3-SAT by up to 22.5 points, SatMAE by up to 8.8), with opposite outcomes even for two MAEs pretrained on the same fMoW-RGB imagery. These effects are robust to patch-token pooling and attentive probing, concentrate on small man-made objects, and are largest with 1-5% labels. Across encoders, the silhouette score of frozen embeddings predicts k-NN accuracy and retrieval mAP almost perfectly (Spearman $\rho \geq 0.96$ on four of five datasets). Pretraining on satellite imagery is thus not a reliable selection criterion; candidate checkpoints must be validated on target data.
| Comments: | Accepted at the 3rd Workshop on Computer Vision for Geospatial Image Analysis (GeoCV) at WACV 2027 |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.13849 [cs.CV] |
| (or arXiv:2611.13849v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.13849 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Rakshith Sathish [view email]
[v1]
Tue, 10 Nov 2026 10:10:29 UTC (1,016 KB)
来源：arXiv cs.AI · arxiv.org