九大卷积神经网络 ( CNN ) 的 PyTorch 实现

下图来源：https://github.com/hoya012/deep_learning_object_detection

52个目标检测模型

---

以下内容来源：https://github.com/shanglianlm0525/PyTorch-Networks

典型网络

典型的卷积神经网络包括：AlexNet、VGG、ResNet; InceptionV1、InceptionV2、InceptionV3、InceptionV4、Inception-ResNet。

• AlexNet: ImageNet Classification with Deep Convolutional Neural Networks, Alex Krizhevsky, 2012

• VGG: Very Deep Convolutional Networks for Large-Scale Image Recognition,Karen Simonyan,2014

• ResNet: Deep Residual Learning for Image Recognition, He-Kaiming, 2015

• InceptionV1: Going deeper with convolutions , Christian Szegedy , 2014

• InceptionV2 and InceptionV3: Rethinking the Inception Architecture for Computer Vision , Christian Szegedy ,2015

• InceptionV4 and Inception-ResNet: Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning , Christian Szegedy ,2016

• DenseNet: Densely Connected Convolutional Networks, 2017

• ResNeXt: Aggregated Residual Transformations for Deep Neural Networks,2017

---

轻量级网络

轻量级网络包括：GhostNet、MobileNets、MobileNetV2、MobileNetV3、ShuffleNet、ShuffleNet V2、SqueezeNet Xception MixNet GhostNet。

• MobileNets: MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

• MobileNetV2: Inverted Residuals and Linear Bottlenecks

• MobileNetV3：Searching for MobileNetV3

• ShuffleNet: An Extremely Efficient Convolutional Neural Network for Mobile Devices

• ShuffleNet V2: Practical Guidelines for Efficient CNN Architecture Design

• SqueezeNet：AlexNet-level accuracy with 50x fewer parameters and < 0.5MB Model Size

• Xception: Deep Learning with Depthwise Separable Convolutions

• MixNet: Mixed Depthwise Convolutional Kernels

---

目标检测网络

目标检测网络包括：SSD、YOLO、YOLOv2、YOLOv3、FCOS、FPN、RetinaNet Objects as Points、FSAF、CenterNet FoveaBox。

• SSD: Single Shot MultiBox Detector,2016

  • 论文地址：https://arxiv.org/pdf/1512.02325.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100530275

• YOLO：You Only Look Once: Unified, Real-Time Object Detection, 2016

  • 论文地址：https://arxiv.org/pdf/1506.02640.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100904605

• YOLOv2: Better, Faster, Stronger,2017

  • 论文地址：https://arxiv.org/pdf/1804.02767.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100904645

• YOLOv3: An Incremental Improvement, 2018

  • 论文地址：https://arxiv.org/pdf/1612.08242.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100904663

• FCOS: Fully Convolutional One-Stage Object Detection, 2019

  • 论文地址：https://arxiv.org/pdf/1904.01355.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/89007219

• FPN：Feature Pyramid Networks for Object Detection, 2017

  • 论文地址：https://arxiv.org/pdf/1612.03144v2.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100864158

• RetinaNet：Focal Loss For Dense Objective Detection

  • 论文地址：https://arxiv.org/pdf/1708.02002.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/102135318

• Objects as Points: Objects as Points,2019

  • 论文地址：https://arxiv.org/pdf/1904.07850v1.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100867545

• FSAF: Feature Selective Anchor-Free Module for Single-Shot Object Detection, 2019

  • 论文地址：https://arxiv.org/pdf/1903.00621.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100942317

• CenterNet: Keypoint Triplets for Object Detection, 2019

  • 论文地址： https://arxiv.org/pdf/1904.08189.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100942259

• FoveaBox: Beyond Anchor-based Object Detector, 2019

  • 论文地址：https://arxiv.org/pdf/1904.03797v1.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100941880

---

语义分割网络

语义分割网络包括：FCN、Fast-SCNN、LEDNet、LRNNet、FisheyeMODNet。

• FCN: Fully Convolutional Networks for Semantic Segmentation

  • 论文地址https://arxiv.org/pdf/1411.4038.pdf

• Fast-SCNN: Fast Semantic Segmentation Network

  • 论文地址：https://arxiv.org/pdf/1902.04502.pdf

• LEDNet: A Lightweight Encoder-Decoder Network for Real-time Semantic Segmentation

  • ；论文地址：https://arxiv.org/pdf/1905.02423.pdf

• LRNNet: A Light-Weighted Network with Efficient Reduced Non-Local Operation for Real-Time Semantic Segmentation

  • 论文地址：https://arxiv.org/pdf/2006.02706.pdf

• FisheyeMODNet: Moving Object detection on Surround-view Cameras for Autonomous Driving (2019)

  • 论文地址：https://arxiv.org/pdf/1908.11789v1.pdf

---

实例分割网络

实例分割网络包括：PolarMask。
PolarMask: Single Shot Instance Segmentation with Polar Representation ,2019

• 论文地址：https://arxiv.org/pdf/1909.13226.pdf

• 论文解读：https://liumin.blog.csdn.net/article/details/101975085

---

人脸检测和识别网络

人脸检测和识别网络包括：FaceBoxes、LFFD、VarGFaceNet。

• FaceBoxes: A CPU Real-time Face Detector with High Accuracy,2018

  • 论文地址：https://arxiv.org/pdf/1708.05234.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/97698853

• LFFD: A Light and Fast Face Detector for Edge Devices,2019

  • 论文地址：https://arxiv.org/pdf/1904.10633.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/100181190

---

人体姿态识别网络

人体姿态识别网络包括：Stacked Hourglass、Networks Simple Baselines、LPN。
StackedHG: Stacked Hourglass Networks for Human Pose Estimation ,2016

• 论文地址：https://arxiv.org/pdf/1603.06937.pdf

• 论文解读：https://liumin.blog.csdn.net/article/details/101484455

Simple Baselines：Simple Baselines for Human Pose Estimation and Tracking

• 论文地址：https://arxiv.org/pdf/1804.06208.pdf

• 论文解读：https://liumin.blog.csdn.net/article/details/103447040

LPN: Simple and Lightweight Human Pose Estimation

• 论文地址：https://arxiv.org/pdf/1911.10346v1.pdf

• 论文解读：https://liumin.blog.csdn.net/article/details/103448034

---

注意力机制网络

注意力机制网络包括：SE Net、scSE、NL Net、GCNet、CBAM。

• SE Net：Squeeze-and-Excitation Networks,2017

  • 论文地址：https://arxiv.org/pdf/1709.01507.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/104370739

• scSE：Concurrent Spatial and Channel Squeeze & Excitation in Fully Convolutional Networks, 2018

  • 论文地址：https://arxiv.org/pdf/1803.02579v2.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/104371065

• NL Net：Non-Local neural networks,2018

  • 论文地址：https://arxiv.org/pdf/1711.07971.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/104371212

• GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond, 2019

  • 论文地址：https://arxiv.org/pdf/1904.11492.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/104375585

• CBAM: Convolutional Block Attention Module, 2018

  • 论文地址：https://arxiv.org/pdf/1807.06521.pdf

  • 论文解读：https://liumin.blog.csdn.net/article/details/104371273

---

人像分割网络

人像分割网络包括：SINet。

• SINet：Extreme Lightweight Portrait Segmentation Networks

  • 论文地址：https://arxiv.org/pdf/1911.09099.pdf

  • 论文解读：https://blog.csdn.net/shanglianlm/article/details/103931852