Light-Weight RetinaNet for Object Detection on Edge Devices
Light-Weight RetinaNet for Object Detection on Edge Devices
复制标题
DOI：
10.1109/wf-iot48130.2020.9221150
复制
发表时间：
2020-06
期刊：
影响因子：
--
通讯作者：
Yixing Li;A. Dua;Fengbo Ren
中科院分区：
文献类型：
--
作者：
Yixing Li;A. Dua;Fengbo Ren
本文旨在减少 Retinanet（一个 mAP-30 层网络）的计算量，以促进其在边缘设备上的实际部署，以提供基于物联网的对象检测服务。我们首先验证 RetinaNet 在所有 mAP-30 层网络中具有最佳的 FLOP-mAP 权衡。然后，我们提出了一种轻量级的 RetinaNet 结构，通过仅减少计算密集型层中的 FLOP 来实现有效的计算精度权衡。与权衡计算与精度输入图像缩放的最常见方法相比，所提出的解决方案显示出一致更好的 FLOPs-mAP 权衡曲线。与原始 RetinaNet 相比，轻量级 RetinaNet 在 1.8 倍的 FLOPs 减少点上实现了 0.3% 的 mAP 改进，并且在边缘计算环境中在英特尔 Arria 10 FPGA 加速器上获得了 1.8 倍的能效。所提出的方法可能可以帮助广泛的对象检测应用程序靠近首选角落，以获得更好的运行时间和准确性，同时在边缘享受更节能的推理。
This paper aims at reducing computation for Retinanet, an mAP-30-tier network, to facilitate its practical deployment on edge devices for providing IoT-based object detection services. We first validate RetinaNet has the best FLOP-mAP trade-off among all mAP-30-tier network. Then, we propose a light-weight RetinaNet structure with effective computation- accuracy trade-off by only reducing FLOPs in computationally intensive layers. Compared with the most common way of trading off computation with accuracy-input image scaling, the proposed solution shows a consistently better FLOPs-mAP trade-off curve. Light-weight RetinaNet achieves a 0.3% mAP improvement at 1.8x FLOPs reduction point over the original RetinaNet, and gains 1.8x more energy-efficiency on an Intel Arria 10 FPGA accelerator in the context of edge computing. The proposed method potentially can help a wide range of the object detection applications to move closer to a preferred corner for a better runtime and accuracy, while enjoys more energy-efficient inference at the edge.