解决小目标检测中高层特征训练缺失问题
CEM-FBGTinyDet: Context-Enhanced Foreground Balance with Gradient Tuning for tiny Objects
- 通过双分支结构融合全局与局部特征增强低层表示
- 动态生成空间掩码,突出判别性区域提升定位精度
- 自适应梯度平衡损失,缓解多尺度目标训练不均
小目标检测暴露了特征金字塔网络的一个根本缺陷:在标准标签分配下,高层特征(P5-P6)常因无正样本锚点而无法参与损失计算,导致其语义表征未被训练。这引发双重问题:(1) 高层特征因缺乏梯度更新成为语义死区;(2) 低层特征缺乏必要语义上下文,难以实现鲁棒分类。为此,我们提出E-FPN-BS,一种融合多尺度特征增强与自适应优化的新型架构。首先,上下文增强模块(CEM)采用双分支处理,对齐并压缩高层特征以实现有效全局-局部融合;其次,前景-背景分离模块(FBSM)生成空间门控掩码,动态增强判别区域。为缓解不同尺度目标间的梯度不平衡,进一步提出动态梯度平衡损失(DCLoss),通过尺度感知的梯度均衡自动调节损失贡献。在多个基准数据集上的大量实验表明,该方法具有卓越性能与强泛化能力。
原文摘要 · Abstract (English)
Tiny object detection (TOD) reveals a fundamental flaw in feature pyramid networks: high-level features (P5-P6) frequently receive zero positive anchors under standard label assignment protocols, leaving their semantic representations untrained due to exclusion from loss computation. This creates dual deficiencies: (1) Stranded high-level features become semantic dead-ends without gradient updates, while (2) low-level features lack essential semantic context for robust classification. We propose E-FPN-BS that systematically converts wasted high-level semantics into low-level feature enhancements. To address these issues, we propose E-FPN-BS, a novel architecture integrating multi-scale feature enhancement and adaptive optimization. First, our Context Enhancement Module(CEM) employs dual-branch processing to align and compress high-level features for effective global-local fusion. Second, the Foreground-Background Separation Module (FBSM) generates spatial gating masks that dynamically amplify discriminative regions. To address gradient imbalance across object scales, we further propose a Dynamic Gradient-Balanced Loss (DCLoss) that automatically modulates loss contributions via scale-aware gradient equilibrium. Extensive experiments across multiple benchmark datasets demonstrate the outstanding performance and generalization ability of our approach.
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。