arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2008.07294cs.CV
收藏

AP-Loss 用于准确的一阶段目标检测

AP-Loss for Accurate One-Stage Object Detection

Shanghai Jiao Tong University(上海交通大学);Intel Labs(英特尔实验室);Multimedia University(多媒体大学);Tencent YouTu Lab(腾讯优图实验室)

Kean Chen, Weiyao Lin, Jianguo Li, John See, Ji Wang, Junni Zou

AI总结:

本文提出基于AP-loss的排序任务替代传统分类任务,通过改进的优化算法解决一阶段目标检测中的类别不平衡问题,提升检测性能。

摘要:

一阶段目标检测器通过同时优化分类损失和定位损失进行训练,其中前者由于大量锚点的存在而遭受严重的前景-背景类别不平衡问题。本文通过提出一个新框架,将一阶段检测器中的分类任务替换为排序任务,并采用平均精度损失(AP-loss)来解决排序问题。由于其非可微性和非凸性,AP-loss不能直接优化。为此,我们开发了一种新的优化算法,该算法无缝结合了感知机学习中的误差驱动更新方案和深度网络中的反向传播算法。我们对所提出算法的良好收敛性和计算复杂性进行了深入分析,理论和实验均表明其优越性。实验结果表明,在目标检测中对现有AP-based优化算法的不平衡问题有显著改进。基于AP-loss的一阶段检测器在各种标准基准上实现了改进的最先进性能。所提出的框架在适应不同网络架构方面也具有高度的通用性。代码可在https://github.com/cccorn/AP-loss获取。

英文摘要:

One-stage object detectors are trained by optimizing classification-loss and localization-loss simultaneously, with the former suffering much from extreme foreground-background class imbalance issue due to the large number of anchors. This paper alleviates this issue by proposing a novel framework to replace the classification task in one-stage detectors with a ranking task, and adopting the Average-Precision loss (AP-loss) for the ranking problem. Due to its non-differentiability and non-convexity, the AP-loss cannot be optimized directly. For this purpose, we develop a novel optimization algorithm, which seamlessly combines the error-driven update scheme in perceptron learning and backpropagation algorithm in deep networks. We provide in-depth analyses on the good convergence property and computational complexity of the proposed algorithm, both theoretically and empirically. Experimental results demonstrate notable improvement in addressing the imbalance issue in object detection over existing AP-based optimization algorithms. An improved state-of-the-art performance is achieved in one-stage detectors based on AP-loss over detectors using classification-losses on various standard benchmarks. The proposed framework is also highly versatile in accommodating different network architectures. Code is available at https://github.com/cccorn/AP-loss .

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑