Conformal Test Martingale-Based Change-Point Detection for Geospatial Object Detectors

Unsupervised domain adaptation for object detectors addresses the problem of improving cross-domain robustness detection from label-rich to label-poor domains, which has been explored in many studies. However, one important issue terms when apply algorithm geospatial not fully considered literature. In this paper, we tackle detecting moment or change-point images changes based on conformal test martingale. Beyond simple introduction martingale-based process, also propose a novel transformation approach original martingale make more efficient. The experiments are conducted with two partitions our released large-scale remote sensing dataset and experimental results empirically demonstrate promising effectiveness efficiency proposed algorithms detection.