Scalable Model-Assisted Multi-Target Estimation in Large Image Collections
For scientists and practitioners needing statistically rigorous population-level estimates from large image collections, this paper provides a scalable framework for multi-target estimation, addressing a key limitation of prior single-scalar approaches.
This work extends Monte Carlo estimation from single-scalar to multi-target estimation in large image collections, combining model predictions with limited human labels. Evaluations on five datasets with 7-80 classes show that importance sampling works best with moderate budgets or few targets, while uniform sampling with control variates excels for many targets or minimal labels, and a subset-based ratio estimator is competitive across all regimes.
Computer vision models are increasingly used as measurement tools to estimate population-level quantities from large image collections, but prediction errors introduce bias and the resulting estimates lack statistical guarantees required in scientific applications. Prior work uses a Monte Carlo framework to combine model predictions with ground-truth annotations by sampling some images for humans to label and is able to provide unbiased estimates with controllable accuracy, but primarily addresses single-scalar estimation. We study the more general problem of multi-target estimation, where many quantities (e.g., class counts or proportions) must be estimated simultaneously, and adapt sampling and estimation strategies from survey sampling to this setting. Evaluations on five detection and segmentation datasets with 7-80 classes show that importance sampling excels with moderate annotation budgets or fewer targets, whereas uniform sampling with control variates is superior when estimating many targets or operating with minimal labels. Additionally, a subset-based ratio estimator remains highly competitive across all regimes. Ultimately, our framework effectively combines biased model predictions and limited human labels into rigorous scientific measurements.