Simple Supervision Is Hard to Beat: A Bitter Lesson from Sparse Target Labels in Domain-Adaptive Object Detection

2026-06-29

AI Summary

This work addresses the challenge of effectively leveraging a small number of labeled target-domain images in unsupervised domain adaptive object detection. The authors propose Random Target Supervision Mixing (RTSM), a method that operates within a teacher-student self-training framework. RTSM jointly optimizes sparse ground-truth annotations through a supervised loss alongside unlabeled branches and incorporates target-domain labels via a random sampling strategy. Extensive experiments demonstrate that RTSM consistently outperforms purely unsupervised domain adaptation approaches across various detectors and tasks, achieving AP50 gains of 1.7–18.3. Notably, the results reveal that under sparse labeling conditions, straightforward supervised strategies can surpass more complex pseudo-labeling feedback mechanisms, establishing RTSM as a strong and effective new baseline for this setting.

Abstract

Source-free domain adaptive object detection adapts a source-trained detector to an unlabeled target domain, typically through teacher-student self-training with pseudo-labels. We revisit this setting when a small, uniformly sampled subset of target images is labeled. We introduce Random-Target Supervised Mixing (RTSM), a simple anchor that incorporates these annotations through a supervised detection loss while leaving the original unlabeled adaptation branch unchanged. Across evaluations spanning four SFDA-OD methods, two object detectors, multiple adaptation tasks, and target-label budgets from 1% to 10%, RTSM consistently improves pure SFDA by 1.7 to 18.3 AP50. We then examine whether the same annotations can provide further gains by steering unlabeled self-training. To this end, we evaluate ten sparse-label feedback plugins covering pseudo-label selection, object completion, and optimization control, which yield limited and method-dependent gains over RTSM. These results reveal a bitter lesson for sparse-label SFDA-OD: simple supervision is hard to beat. RTSM therefore provides a simple yet effective anchor for sparse-label SFDA-OD.

Problem

• source-free domain adaptation
• object detection
• sparse labels
• semi-supervised learning
• pseudo-labeling

Innovation

• source-free domain adaptation
• object detection
• sparse supervision
• pseudo-labeling
• supervised mixing