aiqbench analysis
This study challenges the universal applicability of AI scaling laws by demonstrating that in data-scarce Earth observation tasks, smaller models paired with higher input resolution significantly outperform larger counterparts. The research provides a critical efficiency inversion for practitioners deploying computer vision systems on constrained hardware or limited datasets, showing that model size is not always the primary driver of performance.
When Bigger is Worse: A Practitioner's Guide to Model Selection Under Data Scarcity
arXiv:2603.02142v2 Announce Type: replace-cross Abstract: 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.
Abstract: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.
Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
arXiv:2603.02142 [cs.CV]
(or arXiv:2603.02142v2 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2603.02142 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 286 words
Related coverage
How Do You Choose Your AI Component? An Interview Study of Secure AI Integration in Practice
arXiv · Jul 21, 2026
SelectArena: A Modern Benchmark for Feature Selection on Real-world Tabular Data
arXiv · Oct 12, 2026
Look Before You Lift: Visual and Quantitative Diagnostics for Topological Deep Learning
arXiv · Aug 19, 2026
KUPAS MASTER: Distilling the Tacit Expertise of Master Practitioners into Agent-Ready Experience Corpora
arXiv · Oct 1, 2026