Active-SAOOD actively picks instance-level sparse samples using orientation, classification, localization uncertainty and inter/intra-class diversity, delivering a 9% performance gain at 1% annotation ratio over baseline SAOOD methods.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Active-SAOOD: Active Sparsely Annotated Oriented Object Detection in Remote Sensing Images
Active-SAOOD actively picks instance-level sparse samples using orientation, classification, localization uncertainty and inter/intra-class diversity, delivering a 9% performance gain at 1% annotation ratio over baseline SAOOD methods.