自己整理的深度学习图像和无人驾驶的相关名词深度学习图像无人驾驶

71 阅读 0 评论 47 点赞

我是靠谱客的博主凶狠钢笔，最近开发中收集的这篇文章主要介绍自己整理的深度学习图像和无人驾驶的相关名词深度学习图像无人驾驶，觉得挺不错的，现在分享给大家，希望可以做个参考。

概述

深度学习图像

基础名词

MLP:Multi-Layer Perceptron 多层感知器。
BP:errorBackPropagation BP算法是由学习过程由信号的正向传播与误差的反向传播两个过程组成。
CNN:Convolutional Neural NetWork 卷积神经网络
RNN:Recurrent Neural Network 循环神经网络
LSTM:Long Short-Term Memory 长短期记忆网络

卷积相关

Dilated Convolutions:空洞卷积(DEEPLAB、SSD)
DW:Depthwise Separable Convolution 深度可分离卷积(XCEPTION)
Inverted Residuals: Inverted residuals，通常的residuals block是先经过一个1*1的Conv layer，把feature map的通道数“压”下来，再经过3*3 Conv layer，最后经过一个1*1 的Conv layer，将feature map 通道数再“扩张”回去。即先“压缩”，最后“扩张”回去。 inverted residuals就是先“扩张”，最后“压缩”(MOBILENETV2)

基础网络

RESNET:Deep residual network, ResNet 深度残差网络
XCEPTION:Deep Learning with Depthwise Separable Convolutions
DENSENET:Densely Connected Convolutional Networks
MobileNetV2: MobileNetV2 Inverted Residuals and Linear Bottlenecks
ShuffleNet:ShuffleNet An Extremely Efficient Convolutional Neural Network for Mobile Devices

目标检测、语义分割网络：

SPP:spatial pyramid pooling
ASPP:atrous spatial pyramid pooling（DEEPLAB）
FPN:feature pyramid networks(YOLOV3等)
Region Proposal Networks(FASTER RCNN)

目标检测的算法演变：RCNN->FAST RCNN->FASTER RCNN、YOLOV1->YOLOV2->YOLOV3、SSD
RCNN:Region-based Convolutional Network
YOLO:You Only Look Once:Unified, Real-Time Object Detection PS:以后起名字就要这么任性！~
SSD:Single Shot MultiBox Detector

FCN：Fully Convolutional Networks for Semantic Segmentation
DeepLab:SEMANTIC IMAGE SEGMENTATION WITH DEEP CONVOLUTIONAL NETS AND FULLY CONNECTED CRFS
SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation
DeepLab v3+:Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation

无人驾驶

SENSORS:camera、radar、lidar、GPS
Preception:Lane Detection(OPENCV)、Traffic Light Detection & Classification、Traffic Light Detection & Classification、Object Detection & Tracking(EKF extended kalman filter)、Free Space Detection (Object Detection). Location(Partical filters?SLAM?)
Planning:route planning、prediction、behavior planning、trajectory planning
Control: PID(proportion integral derivative)、MPC(Model Predictive Control)