动态调整模型更新策略，提升视觉语言检测器部署时的稳定性
DESA-TTA: Dynamic EMA and Source Anchoring for Test-Time Adaptation
- 根据置信度和框密度动态调整教师模型更新系数
- 在VOC-C上相比零样本推理提升14.5点AP₅₀，推理速度提高55%
- 适合需要高鲁棒性且对延迟敏感的实时目标检测场景
视觉语言目标检测器（VLODs）在零样本设置下表现优异，但在部署过程中易受分布偏移影响。现有的均值教师测试时自适应（TTA）方法通过教师生成伪标签来更新学生模型，但对固定的指数移动平均（EMA）系数敏感，且重复优化中噪声伪标签会导致学生模型累积漂移。本文提出动态EMA与源锚定方法（DESA-TTA），通过动态时间平均和源锚定联合调控教师更新与学生漂移。动态时间平均依据伪标签置信度与框密度估计教师不确定性，自适应选择样本级EMA系数；源锚定将更新后的学生参数部分恢复至预训练值，锚定强度随学生漂移增大而增强。在多种分布偏移及两种VLOD架构上实验表明，DESA-TTA显著优于现有TTA方法。在VOC-C数据集上，相比零样本推理提升14.5点AP₅₀，且推理吞吐量比当前最优TTA方法高出55%。代码已开源。
原文摘要 · Abstract (English)
Vision-language object detectors (VLODs) achieve strong zero-shot performance but remain vulnerable to distribution shifts during deployment. Mean-teacher methods for test-time adaptation (TTA) can improve robustness by updating a student model using teacher-generated pseudo-labels. However, mean-teacher TTA is highly sensitive to the choice of a fixed exponential moving average (EMA) coefficient for teacher updates, and repeated optimization with noisy pseudo-labels can cause cumulative student drift. We propose Dynamic EMA and Source Anchoring for TTA (DESA-TTA), a low-overhead method that jointly regulates teacher updates and student drift through dynamic temporal averaging and source anchoring. Dynamic temporal averaging estimates teacher uncertainty from pseudo-label confidence and box density and uses it to select a sample-wise EMA coefficient within bounds determined by teacher parameter drift. Source anchoring partially restores the updated student parameters toward their pretrained values, with the anchoring strength increasing according to student drift. Experiments across diverse distribution shifts and two VLOD architectures show consistent improvements over existing TTA methods. On VOC-C, DESA-TTA improves AP$_{50}$ by 14.5 points over zero-shot inference while achieving 55\% higher inference throughput than the previous state-of-the-art TTA method for YOLO-World. Our code: https://github.com/imatif17/DESA-TTA
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。