小模型高分辨率在数据稀缺时反而更高效，打破规模越大越好的常识。
When Bigger is Worse: A Practitioner's Guide to Model Selection Under Data Scarcity
- 用小模型+高分辨率替代大模型，效率提升22倍
- 小模型在精度上仅略低于大模型，差异小于实验波动
- 适合资源有限的遥感图像检测任务
缩放定律假设更大的模型在更多数据上训练会持续优于小模型，这一假设在计算机视觉中主导模型选择，但在资源受限的地球观测（EO）领域尚未验证。我们在马达加斯加屋顶光伏检测任务中，系统分析了模型大小、数据量和输入分辨率三个维度的缩放效应，共完成180次训练，覆盖60种配置。以单位模型尺寸的mAP₅₀为效率指标，发现存在显著效率反转：YOLO11N在无精度损失情况下实现最高效率（比YOLO11X高22倍），其绝对mAP₅₀为0.459，仅次于YOLO11S，差距小于实验随机波动，直接违背传统缩放假设。分辨率是资源配置的关键杠杆：仅用10%数据但将分辨率从416像素提升至1280像素，其效率增益相当于使用全量低分辨率数据。该结论在所有60组实验中均成立，小模型高分辨率配置在准确率-吞吐量联合空间中始终占优，无需权衡。在数据稀缺的地球观测场景中，缩放假设不仅失效，甚至反转。
原文摘要 · Abstract (English)
Scaling laws assume larger models trained on more data consistently outperform smaller ones -- an assumption that drives model selection in computer vision but remains untested in resource-constrained Earth observation (EO). We conduct a systematic efficiency analysis across three scaling dimensions: model size, dataset size, and input resolution, on rooftop photovoltaic (PV) detection in Madagascar, yielding 180 training runs across 60 configurations. Optimizing for model efficiency (mAP$_{50}$ per unit of model size), we find a consistent efficiency inversion: YOLO11N achieves the highest efficiency ($22\times$ higher than YOLO11X) with no accuracy penalty: it reaches the second highest absolute mAP$_{50}$ (0.459), outperforming all models except YOLO11S by a margin smaller than run-to-run variance, directly contradicting the scaling prior. Resolution is the dominant resource allocation lever: moving from 416 px to 1280 px at 10 % of the data matches the efficiency gain of collecting the full dataset at low resolution. These findings are robust to the deployment objective: small high-resolution configurations are Pareto-dominant across all 60 experimental setups in the joint accuracy-throughput space, leaving no tradeoff to resolve. In data-scarce EO, the scaling prior does not just fail: it inverts.
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。