EfficientDet: Scalable and Efficient Object Detection
Mingxing Tan Ruoming Pang Quoc V. Le Google Research, Brain Team {tanmingxing, rpang,
Abstract
Model efficiency has become increasingly important in computer vision. In this paper, we systematically study neural network architecture design choices for object detection and propose several key optimizations to improve efficiency. First, we propose a weighted bi-directional feature pyramid network (BiFPN), which allows easy and fast multi-scale feature fusion; Second, we propose a compound scaling method that uniformly scales the resolution, depth, and width for all backbone, feature network, and box/class prediction networks at the same time. Based on these optimizations and better backbones, we have developed a new family of object detectors, called EfficientDet, which consistently achieve much better efficiency than prior art across a wide spectrum of resource constraints. In particular, with single-model and single-scale, our EfficientDet-D7 achieves state-of-the-art 55.1 AP on COCO test-dev with 77M parameters and 410B FLOPs111Similar to [14, 39], FLOPs denotes number of multiply-adds., being 4x – 9x smaller and using 13x – 42x fewer FLOPs than previous detectors. Code is available at https://github.com/google/automl/tree/master/efficientdet.
中文速览
目标检测模型越来越准，但也越来越笨重，难以部署到机器人或自动驾驶等资源受限的场景。为此，谷歌团队提出了两项核心改进：一是加权双向特征金字塔网络（BiFPN，Weighted Bi-directional Feature Pyramid Network），通过可学习权重让不同尺度的特征在双向融合时各有侧重，而不是简单地一视同仁地相加；二是复合缩放策略（Compound Scaling），同时按比例调整骨干网络、特征网络和预测头的分辨率、深度与宽度，而非只改动其中一个维度。将这两项改进与高效骨干网络 EfficientNet 结合，作者构建出 EfficientDet 系列检测器，在 COCO 基准上，旗舰模型 EfficientDet-D7 以 7700 万参数和 410B FLOPs 达到 55.1 AP，比此前最佳方案精度高出 4 个点，参数量缩小约 2.7 倍，计算量减少约 7.4 倍。这项工作的意义在于，它系统性地证明了"大力出奇迹"并非唯一路径——通过有原则的架构设计和协同缩放，完全可以在精度和效率之间同时取得突破，为在边缘设备上落地高性能检测提供了切实可行的方案。
原文 arXiv:1911.09070;中英对照 + 大白话阅读 https://aha.fim.ai/paper/1911.09070v7