发表机构
Sam Houston State University（萨姆休斯顿州立大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究提出任务敏感几何，通过梯度导出的通道空间刻画图像退化下目标检测的表示迁移，干预实验挽回13.22%丢失对象，验证了局部方向族依赖的几何结构，但全局训练改进仍待解决。
AI 中文摘要
图像退化下的目标检测可以从干净图像监督中受益，但总体收益并不意味迁移的表示变化是均匀有用的。我们研究干净任务知识如何影响退化图像表示，以及局部对结构化表示方向的响应能否用几何方式刻画。使用配对的干净和高斯退化BDD100K图像，我们表明干净教师蒸馏提高了观测到的检测精度，同时产生了异质的对象级迁移。我们分离出一个与直接干净教师对齐互补的表示成分，并通过正交桥将其映射到蒸馏学生空间。受控干预挽回了蒸馏表示下丢失的13.22%对象，而范数匹配随机扰动仅挽回4.30%，且对保留对象的损害非常低。我们引入任务敏感几何，一种由归一化检测损失梯度构建的梯度导出的通道空间几何。在保留队列上，映射互补方向在该冻结几何中的朝向与局部干预响应幅度呈正相关，在控制干预幅度后（偏Spearman ρ = 0.242，95% CI [0.108, 0.359]）。该关系在独立数据上复现（ρ = 0.180），并在RT-DETR-L上复现（ρ = 0.227），但直接干净教师残差族不成立，且大干预时减弱。基于这些信号的路径规则和专门蒸馏目标未在CLEANKD上产生统计上可靠的增益。这些结果支持一种局部的、方向族依赖的任务敏感几何，同时表明将这种结构转化为改进的全局训练仍是一个开放问题。
英文摘要
Object detection under image degradation can benefit from clean-image supervision, but aggregate gains do not imply that transferred representation changes are uniformly useful. We study how clean task knowledge affects degraded-image representations and whether local responses to structured representation directions can be characterized geometrically. Using paired clean and Gaussian-degraded BDD100K images, we show that clean-teacher distillation improves observed detection accuracy while producing heterogeneous object-level transfer. We isolate a representation component complementary to direct clean-teacher alignment and map it into the distilled student space through an orthogonal bridge. Controlled interventions rescue 13.22% of objects lost under the distilled representation, versus 4.30% under norm-matched random perturbations, with very low harm on preserved objects. We introduce task-sensitive geometry, a gradient-derived channel-space geometry constructed from normalized detection-loss gradients. On a reserved cohort, mapped-complement orientation within this frozen geometry is positively associated with local intervention-response magnitude after controlling for intervention magnitude (partial Spearman $ρ$ = 0.242, 95% CI [0.108, 0.359]). The relationship eplicates on independent data ($ρ$ = 0.180) and with RT-DETR-L ($ρ$ = 0.227), but not for the direct clean-teacher residual family, and it weakens for large interventions. Routing rules and specialized distillation objectives based on these signals do not yield statistically reliable gains over CLEANKD. These results support a local, direction-family-dependent task-sensitive geometry while showing that converting such structure into improved global training remains an open problem.
Comments20 pages, 5 figures, 1 table
