## 摘要

In this paper we describe a new mobile architecture, MobileNetV2, that improves the state of the art performance of mobile models on multiple tasks and benchmarks as well as across a spectrum of different model sizes. We also describe efficient ways of applying these mobile models to object detection in a novel framework we call SSDLite. Additionally, we demonstrate how to build mobile semantic segmentation models through a reduced form of DeepLabv3 which we call Mobile DeepLabv3. is based on an inverted residual structure where the shortcut connections are between the thin bottleneck layers. The intermediate expansion layer uses lightweight depthwise convolutions to filter features as a source of non-linearity. Additionally, we find that it is important to remove non-linearities in the narrow layers in order to maintain representational power. We demonstrate that this improves performance and provide an intuition that led to this design. Finally, our approach allows decoupling of the input/output domains from the expressiveness of the transformation, which provides a convenient framework for further analysis. We measure our performance on ImageNet [1] classification, COCO object detection [2], VOC image segmentation [3]. We evaluate the trade-offs between accuracy, and number of operations measured by multiply-adds (MAdd), as well as actual latency, and the number of parameters.

MobileNetV2基于反向残差结构，在两个很小（通道数很少）的bottleneck层之间执行残差连接。中间扩展层使用轻量级深度卷积来过滤非线性特征。此外，我们发现为了保持表达能力，消除窄层中的非线性操作是很重要的。我们证明这样操作能提高性能，并说明实现这种设计的直觉

## 章节内容

• 首先介绍了MobileNetV2中使用的各个关键模块，包括深度可分离卷积、线性瓶颈层、反向残差块
• 其次介绍了MobileNetV2模型的整体架构和实现
• 最后通过实验证明MobileNetV2在目标识别、检测（提出了SSDLite）、分割方面的进步

## 瓶颈残差块

MobileNetV1主要使用了深度可分离卷积层进行构建，V2在此基础上进一步研究如何更好的传递信息，提出了线性瓶颈层和反向残差块的概念，最后组合成瓶颈残差块（bottleneck residual block

### ReLU6

ReLU6ReLU的输出进行了限制，其最大输出值为6，适用于移动端低精度设备

$ReLU6(x) = min(max(0, x), 6)$

## 模型架构

### 整体架构

• t：扩展率
• c：通道数
• n：重复次数
• s：步长

### stride=1/2

stride=1时，瓶颈残差块使用了残差连接

## 实验

### 训练细节

• 优化器：RMPSProp．衰减和动量大小为0.9
• 权重衰减：4e-5
• 学习率调度：初始学习率为4.5e-3，每轮衰减率0.98
• 批量大小：96
• GPU数目：16

### SSDLite

SSD进行了修改，使用MobileNetV2作为特征提取层，同时将预测层的标准卷积替换为深度可分离卷积，称该变体为SSDLite

## 小结

MobileNetV2最重要的贡献在于实现了一个新的模块 - 瓶颈残差块。其首先将输入的低维压缩信息扩展到高维空间，然后通过一个轻量级深度卷积进行过滤，最后使用线性瓶颈层投影回低维空间