论文标题
动态特征金字塔网络用于对象检测
Dynamic Feature Pyramid Networks for Object Detection
论文作者
论文摘要
特征金字塔网络（FPN）是现代对象检测框架中的关键组件。大多数现有FPN变体的性能增长主要归因于计算负担的增加。试图通过扩展接受场来增强FPN的尝试丰富了空间信息，这有望在很大程度上提高检测准确性。在本文中，我们首先研究了扩大接受场如何影响FPN的准确性和计算成本。我们探索了一个称为Inception FPN的基线模型，其中每个横向连接都包含具有不同内核大小的卷积过滤器。此外，我们指出，并非所有对象都需要如此复杂的计算，并提出了新的动态FPN（DYFPN）。 DYFPN的输出特征将根据动态门控操作使用自适应选择的分支来计算。因此，所提出的方法可以为在准确性和计算成本之间获得更好的权衡提供更有效的动态推断。对MS-Coco基准进行的广泛实验表明，拟议的DYFPN可以通过最佳的计算资源分配显着提高性能。例如，用DYFPN取代Inception FPN可减少其约40％的拖鞋，同时保持相似的高性能。
Feature pyramid network (FPN) is a critical component in modern object detection frameworks. The performance gain in most of the existing FPN variants is mainly attributed to the increase of computational burden. An attempt to enhance the FPN is enriching the spatial information by expanding the receptive fields, which is promising to largely improve the detection accuracy. In this paper, we first investigate how expanding the receptive fields affect the accuracy and computational costs of FPN. We explore a baseline model called inception FPN in which each lateral connection contains convolution filters with different kernel sizes. Moreover, we point out that not all objects need such a complicated calculation and propose a new dynamic FPN (DyFPN). The output features of DyFPN will be calculated by using the adaptively selected branch according to a dynamic gating operation. Therefore, the proposed method can provide a more efficient dynamic inference for achieving a better trade-off between accuracy and computational cost. Extensive experiments conducted on MS-COCO benchmark demonstrate that the proposed DyFPN significantly improves performance with the optimal allocation of computation resources. For instance, replacing inception FPN with DyFPN reduces about 40% of its FLOPs while maintaining similar high performance.
