无需训练即可评估合成数据集质量，提升目标检测效率
Training-Free Metrics for Synthetic Object Detection Data: A Proxy for Detector Performance
- 提出基于元数据分层的特征匹配方法CCDM，按图像内容分组比较
- 在COCO和VisDrone-DET上与YOLOv8 mAP相关性高达0.97和0.96
- 适用于真实标注或伪标签，可快速筛选高质量合成数据
合成图像正被广泛用于补充稀缺的真实目标检测数据。然而，并非所有合成数据集效果相同，而唯一判断其价值的方法是用它们训练检测器，这既耗时又需密集标注。本文探讨是否可通过无训练度量来排序候选合成数据集的下游实用性。现有图像集度量如FID、KID和MMD通过单一全局统计量比较特征分布，但在检测数据选择中存在双重缺陷：忽略每张图像的组成细节（物体数量、框尺度、类别混合），且在固定组成下全局平均会掩盖区分高mAP与低mAP数据的关键外观差异。为此，我们提出条件-组合域匹配（CCDM），将任意特征空间距离转化为基于元数据定义的分层比较，在不训练检测器的前提下对候选集与目标集进行分组匹配。在COCO和VisDrone-DET上，最优的CCDM变体与YOLOv8 mAP排名高度一致（斯皮尔曼相关系数ρ = 0.97 和 0.96），优于FID、KID和MMD。此外，当参考元数据来自检测器伪标签而非真实标注时，CCDM依然有效。
原文摘要 · Abstract (English)
Synthetic images are increasingly used to augment scarce real data for object detection. However, not all synthetic sets help equally, and the only way to know a set's value is to train a detector on it, which is slow and demands dense annotation. We ask whether a training-free metric can instead rank candidate synthetic training sets by their downstream utility. Existing image-set metrics such as FID, KID, and MMD compare two feature distributions with a single global statistic, which we show is mis-specified for detection-data selection in two ways: it is blind to per-image composition (object count, box scale, class mix), and even at fixed composition its global averaging washes out the appearance differences that separate high-mAP pools from low-mAP ones. We propose Conditional-Composition Domain Match (CCDM), which converts any feature-space distance into a composition-stratified comparison, matching candidate and target within metadata-defined strata without training a detector. On COCO and VisDrone-DET, the best CCDM variant ranks 19 candidate training sets in strong agreement with YOLOv8 mAP (Spearman \r{ho} = 0.97 and 0.96), outperforming FID, KID, and MMD. Furthermore, CCDM holds when reference metadata comes from detector pseudo-labels rather than ground-truth boxes.
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。