课程大纲第一课：SLAM概论和架构 1.从机器人的体系结构讨论SLAM的提出和发展 2.滤波器是什么，谁真正的推动了SLAM？ 3.SLAM的新突破-图优化 4.SLAM的完整知识体系结构介绍，基于Linux和ROS进行SLAM的进行本课程学习 5.ROS基础：RGB-D点云示例
第二课：SLAM基本理论一：坐标系、刚体运动和李群 1.SLAM的数学表达 2.欧式坐标系和刚体姿态表示 3.李群和李代数 4.实例：Eigen和Sophus在滤波器上的应用
第三课：SLAM基本理论二：从贝叶斯开始学滤波器
1.随机状态和估计 2.卡尔曼滤波器 3.扩展卡尔曼滤波器和SLAM 4.粒子滤波器和SLAM
5.实例：基于卡尔曼滤波器的SLAM实例
第四课：SLAM基本理论三：图优化
1.从滤波器的痛来谈图优化 2.CovisibilityGraph和最小二乘
3.浅谈Marginlization 4.实例：G2O图优化实战
第五课：SLAM的传感器 1.SLAM传感器综述 2.视觉类传感器（单目、双目和RGBD相机）a.相机模型和标定b.特征提取和匹配 3.主动类传感器--激光a.激光模型和不同激光特性b.激光特征和匹配 4.实例：a.特征提取和立体视觉的深度结算；b.激光数据的基本处理
第六课：视觉里程计和回路检测 1.视觉里程计的综述 2.基于特征法的视觉里程计：PNP 3.基于直接法的视觉里程计：PhotometricError 4.基于立体视觉法的:ICP 5.基于词袋模型的回路检测 6.实例：a.PNP位姿估计b.直接法位姿估计c.回路检测
第七课：激光里程计和回路检测 1.激光里程计简介 2.激光里程计算法LOAM和VLOAM简单介绍 3.激光回路检测的特殊性和主要难点 4.伯克利的BLAM和谷歌Cartographer中回路检测的核心思路介绍 5.实例:LOAM,Cartographer测试
第八课：地图以及无人驾驶系统 1.SLAM中的不同地图系统介绍 2.高精度地图介绍 3.语义地图介绍 4.拓扑地图介绍 5.实例：粒子滤波定位实现
第九课：视觉和无人机、室内辅助导航和AR/VR
1.视觉SLAM的整体重述和实战
2.SLAM、无人机和状态机
3.GoogleTango和盲人导航 4.SLAM的小刺激：AR/VR 5.实例：视觉SLAM的AR实例
第十课：深度学习和SLAM
1.SLAM的过去、现在和未来 2.长航程SLAM的可能性 3.单目深度估计和分割和场景语义 4.动态避障 5.新的特征表达 6.课程总结
5
大数据自动驾驶
End-to-end Learning of Driving Models from Large-scale Video Datasets
https://arxiv.org/abs/1612.01079
记忆 注意力 与 语义
使用spark 运行ros 运行模拟
测试数据集 www.cvlibs.net/datasets/kitti/
knowlege of MASK-RCNN traffic segment reconignzs 原文：https://news.voyage.auto/under-the-hood-of-a-self-driving-car-78e8bbce62a6
FORScan：http://www.forscan.org/Dataspeed
http://blog.csdn.net/AdamShan/article/details/78248421?locationNum=7&fps=1
http://blog.csdn.net/AdamShan/article/details/78265754?locationNum=2&fps=1
https://github.com/udacity/CarND-Extended-Kalman-Filter-Project/tree/master/src
http://blog.csdn.net/lybaihu/article/details/54943545?locationNum=8&fps=1 贝叶斯对抗生成网络论文地址：https://arxiv.org/pdf/1705.09558.pdf https://github.com/davidbrai/deep-learning-traffic-lights 交通灯识别 https://github.com/awjuliani https://github.com/awjuliani/TF-Tutorials
http://blog.csdn.net/weixin_37239947/article/details/74939650处理
tensor2tensor http://blog.csdn.net/shenxiaolu1984/article/details/73736259 http://blog.csdn.net/amds123/article/details/73485914 https://github.com/tensorflow/tensor2tensor
生成对抗文本到图像的合成（Generative Adversarial Text to Image Synthesis） 【代码】https://github.com/paarthneekhara/text-to-image
http://www.ctoutiao.com/172487.html gan 代码
https://github.com/priya-dwivedi udacity 学生代码
http://lib.csdn.net/article/aiframework/60538
http://lib.csdn.net/article/89/61236?knId=1818
<<<<<<< HEAD
https://github.com/priya-dwivedi
Canny 原理 http://www.pclcn.org/study/shownews.php?lang=cn&id=111 http://www.opencv.org.cn/opencvdoc/2.3.2/html/doc/tutorials/imgproc/imgtrans/canny_detector/canny_detector.html?highlight=canny#canny
http://selfdrivingcars.mit.edu/resources
三角剖分的算法比较成熟。目前有很多的库（包括命令行的和GUI的可以用）。
常用的算法叫Delaunay Triangulation，具体算法原理见 http://www.cnblogs.com/soroman/archive/2007/05/17/750430.html
这里收集一些开元的做可以测试三角剖分的库
- Shewchuk的http://www.cs.cmu.edu/~quake/triangle.html，据说效率非常高！
- MeshLab http://www.cs.cmu.edu/~quake/triangle.html，非常易于上手，只要新建工程，读入三维坐标点，用工具里面的Delaunay Trianglulation来可视化就好了。而且它是开源的！具体教程去网站上找吧。
- Qhull http://www.qhull.org/
- PCL库，http://pointclouds.org/documentation/tutorials/greedy_projection.php
无序点云快速三角化
http://www.pclcn.org/study/shownews.php?lang=cn&id=111
opencv 基本操作 https://segmentfault.com/a/1190000003742422
cv 到cv2的不同 http://www.aiuxian.com/article/p-395730.html
已经fork https://github.com/mbeyeler/opencv-machine-learning
前言
https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/00.00-Preface.ipynb
机器学习的味道
在OpenCV中使用数据
使用Python的NumPy软件包处理数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.01-Dealing-with-Data-Using-Python-NumPy.ipynb 在Python中加载外部数据集 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.02-Loading-External-Datasets-in-Python.ipynb 使用Matplotlib可视化数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.03-Visualizing-Data-Using-Matplotlib.ipynb 使用OpenCV的TrainData容器处理数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.05-Dealing-with-Data-Using-the-OpenCV-TrainData-Container-in-C%2B%2B.ipynb 监督学习的第一步
用评分功能测量模型性能 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.01-Measuring-Model-Performance-with-Scoring-Functions.ipynb 了解k-NN算法 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.02-Understanding-the-k-NN-Algorithm.ipynb 使用回归模型预测持续成果 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.03-Using-Regression-Models-to-Predict-Continuous-Outcomes.ipynb 应用拉索和岭回归 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.04-Applying-Lasso-and-Ridge-Regression.ipynb 使用Logistic回归分类虹膜物种 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.05-Classifying-Iris-Species-Using-Logistic-Regression.ipynb 代表数据和工程特性
预处理数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.01-Preprocessing-Data.ipynb 减少数据的维度 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.02-Reducing-the-Dimensionality-of-the-Data.ipynb 代表分类变量 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.03-Representing-Categorical-Variables.ipynb 表示文本特征 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.04-Represening-Text-Features.ipynb 代表图像 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.05-Representing-Images.ipynb 使用决策树进行医学诊断
建立你的第一决策树 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/05.01-Building-Your-First-Decision-Tree.ipynb 使用决策树诊断乳腺癌 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/05.02-Using-Decision-Trees-to-Diagnose-Breast-Cancer.ipynb 使用决策树回归 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/05.03-Using-Decision-Trees-for-Regression.ipynb 用支持向量机检测行人
实施您的第一支持向量机 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/06.01-Implementing-Your-First-Support-Vector-Machine.ipynb 检测野外行人 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/06.02-Detecting-Pedestrians-in-the-Wild.ipynb 附加SVM练习 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/06.03-Additional-SVM-Exercises.ipynb 用贝叶斯学习实现垃圾邮件过滤器
实现我们的第一个贝叶斯分类器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/07.01-Implementing-Our-First-Bayesian-Classifier.ipynb 分类电子邮件使用朴素贝叶斯 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/07.02-Classifying-Emails-Using-Naive-Bayes.ipynb 用无监督学习发现隐藏的结构
了解k均值聚类 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.01-Understanding-k-Means-Clustering.ipynb 使用k-Means压缩彩色图像 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.02-Compressing-Color-Images-Using-k-Means.ipynb 使用k-Means分类手写数字 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.03-Classifying-Handwritten-Digits-Using-k-Means.ipynb 实施聚集层次聚类 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.04-Implementing-Agglomerative-Hierarchical-Clustering.ipynb 使用深度学习分类手写数字
了解感知器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.01-Understanding-Perceptrons.ipynb 在OpenCV中实现多层感知器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.02-Implementing-a-Multi-Layer-Perceptron-in-OpenCV.ipynb 认识深度学习 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.03-Getting-Acquainted-with-Deep-Learning.ipynb 在OpenCV中培训MLP以分类手写数字 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.04-Training-an-MLP-in-OpenCV-to-Classify-Handwritten-Digits.ipynb 训练深层神经网络使用Keras分类手写数字 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.05-Training-a-Deep-Neural-Net-to-Classify-Handwritten-Digits-Using-Keras.ipynb 将不同的算法合并成一个合奏
了解组合方法 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.01-Understanding-Ensemble-Methods.ipynb 将决策树组合成随机森林 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.02-Combining-Decision-Trees-Into-a-Random-Forest.ipynb 使用随机森林进行人脸识别 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.03-Using-Random-Forests-for-Face-Recognition.ipynb 实施AdaBoost https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.04-Implementing-AdaBoost.ipynb 将不同的模型组合成投票分类器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.05-Combining-Different-Models-Into-a-Voting-Classifier.ipynb 使用超参数调整选择正确的模型
评估模型 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.01-Evaluating-a-Model.ipynb 了解交叉验证，Bootstrapping和McNemar的测试 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.02-Understanding-Cross-Validation-Bootstrapping-and-McNemar's-Test.ipynb 使用网格搜索调整超参数 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.03-Tuning-Hyperparameters-with-Grid-Search.ipynb 链接算法一起形成管道 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.04-Chaining-Algorithms-Together-to-Form-a-Pipeline.ipynb 结束语
https://github.com/tensorflow/models/tree/master/object_detection mobilenet
https://research.googleblog.com/2017/06/mobilenets-open-source-models-for.html
带有MobileNets的SSD(Single Shot Multibox Detector)
带有Inception V2的SSD
带有Resnet 101的R-FCN（Region-based Fully Convolutional Networks）
带有Resnet 101的 Faster RCNN
带有Inception Resnet v2的Faster RCNN
https://github.com/tensorflow/tensorflow/commit/055500bbcea60513c0160d213a10a7055f079312
mobil net https://github.com/tensorflow/models/tree/master/inception 准备数据 https://github.com/zehaos/MobileNet
https://github.com/balancap/SSD-Tensorflow
2017.9
https://github.com/udacity/CarND-Term1-Starter-Kit 环境配置
http://blog.csdn.net/xukai871105/article/details/39255089 树莓派mqtt
https://github.com/priya-dwivedi
Canny 原理 http://www.pclcn.org/study/shownews.php?lang=cn&id=111 http://www.opencv.org.cn/opencvdoc/2.3.2/html/doc/tutorials/imgproc/imgtrans/canny_detector/canny_detector.html?highlight=canny#canny
http://selfdrivingcars.mit.edu/resources
三角剖分的算法比较成熟。目前有很多的库（包括命令行的和GUI的可以用）。
常用的算法叫Delaunay Triangulation，具体算法原理见 http://www.cnblogs.com/soroman/archive/2007/05/17/750430.html
这里收集一些开元的做可以测试三角剖分的库
- Shewchuk的http://www.cs.cmu.edu/~quake/triangle.html，据说效率非常高！
- MeshLab http://www.cs.cmu.edu/~quake/triangle.html，非常易于上手，只要新建工程，读入三维坐标点，用工具里面的Delaunay Trianglulation来可视化就好了。而且它是开源的！具体教程去网站上找吧。
- Qhull http://www.qhull.org/
- PCL库，http://pointclouds.org/documentation/tutorials/greedy_projection.php
无序点云快速三角化
http://www.pclcn.org/study/shownews.php?lang=cn&id=111
opencv 基本操作 https://segmentfault.com/a/1190000003742422
cv 到cv2的不同 http://www.aiuxian.com/article/p-395730.html
已经fork https://github.com/mbeyeler/opencv-machine-learning
前言
https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/00.00-Preface.ipynb
机器学习的味道
在OpenCV中使用数据
使用Python的NumPy软件包处理数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.01-Dealing-with-Data-Using-Python-NumPy.ipynb 在Python中加载外部数据集 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.02-Loading-External-Datasets-in-Python.ipynb 使用Matplotlib可视化数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.03-Visualizing-Data-Using-Matplotlib.ipynb 使用OpenCV的TrainData容器处理数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/02.05-Dealing-with-Data-Using-the-OpenCV-TrainData-Container-in-C%2B%2B.ipynb 监督学习的第一步
用评分功能测量模型性能 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.01-Measuring-Model-Performance-with-Scoring-Functions.ipynb 了解k-NN算法 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.02-Understanding-the-k-NN-Algorithm.ipynb 使用回归模型预测持续成果 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.03-Using-Regression-Models-to-Predict-Continuous-Outcomes.ipynb 应用拉索和岭回归 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.04-Applying-Lasso-and-Ridge-Regression.ipynb 使用Logistic回归分类虹膜物种 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/03.05-Classifying-Iris-Species-Using-Logistic-Regression.ipynb 代表数据和工程特性
预处理数据 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.01-Preprocessing-Data.ipynb 减少数据的维度 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.02-Reducing-the-Dimensionality-of-the-Data.ipynb 代表分类变量 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.03-Representing-Categorical-Variables.ipynb 表示文本特征 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.04-Represening-Text-Features.ipynb 代表图像 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/04.05-Representing-Images.ipynb 使用决策树进行医学诊断
建立你的第一决策树 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/05.01-Building-Your-First-Decision-Tree.ipynb 使用决策树诊断乳腺癌 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/05.02-Using-Decision-Trees-to-Diagnose-Breast-Cancer.ipynb 使用决策树回归 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/05.03-Using-Decision-Trees-for-Regression.ipynb 用支持向量机检测行人
实施您的第一支持向量机 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/06.01-Implementing-Your-First-Support-Vector-Machine.ipynb 检测野外行人 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/06.02-Detecting-Pedestrians-in-the-Wild.ipynb 附加SVM练习 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/06.03-Additional-SVM-Exercises.ipynb 用贝叶斯学习实现垃圾邮件过滤器
实现我们的第一个贝叶斯分类器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/07.01-Implementing-Our-First-Bayesian-Classifier.ipynb 分类电子邮件使用朴素贝叶斯 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/07.02-Classifying-Emails-Using-Naive-Bayes.ipynb 用无监督学习发现隐藏的结构
了解k均值聚类 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.01-Understanding-k-Means-Clustering.ipynb 使用k-Means压缩彩色图像 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.02-Compressing-Color-Images-Using-k-Means.ipynb 使用k-Means分类手写数字 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.03-Classifying-Handwritten-Digits-Using-k-Means.ipynb 实施聚集层次聚类 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/08.04-Implementing-Agglomerative-Hierarchical-Clustering.ipynb 使用深度学习分类手写数字
了解感知器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.01-Understanding-Perceptrons.ipynb 在OpenCV中实现多层感知器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.02-Implementing-a-Multi-Layer-Perceptron-in-OpenCV.ipynb 认识深度学习 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.03-Getting-Acquainted-with-Deep-Learning.ipynb 在OpenCV中培训MLP以分类手写数字 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.04-Training-an-MLP-in-OpenCV-to-Classify-Handwritten-Digits.ipynb 训练深层神经网络使用Keras分类手写数字 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/09.05-Training-a-Deep-Neural-Net-to-Classify-Handwritten-Digits-Using-Keras.ipynb 将不同的算法合并成一个合奏
了解组合方法 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.01-Understanding-Ensemble-Methods.ipynb 将决策树组合成随机森林 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.02-Combining-Decision-Trees-Into-a-Random-Forest.ipynb 使用随机森林进行人脸识别 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.03-Using-Random-Forests-for-Face-Recognition.ipynb 实施AdaBoost https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.04-Implementing-AdaBoost.ipynb 将不同的模型组合成投票分类器 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/10.05-Combining-Different-Models-Into-a-Voting-Classifier.ipynb 使用超参数调整选择正确的模型
评估模型 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.01-Evaluating-a-Model.ipynb 了解交叉验证，Bootstrapping和McNemar的测试 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.02-Understanding-Cross-Validation-Bootstrapping-and-McNemar's-Test.ipynb 使用网格搜索调整超参数 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.03-Tuning-Hyperparameters-with-Grid-Search.ipynb 链接算法一起形成管道 https://github.com/mbeyeler/opencv-machine-learning/blob/master/notebooks/11.04-Chaining-Algorithms-Together-to-Form-a-Pipeline.ipynb 结束语
https://github.com/tensorflow/models/tree/master/object_detection mobilenet
https://research.googleblog.com/2017/06/mobilenets-open-source-models-for.html
带有MobileNets的SSD(Single Shot Multibox Detector)
带有Inception V2的SSD
带有Resnet 101的R-FCN（Region-based Fully Convolutional Networks）
带有Resnet 101的 Faster RCNN
带有Inception Resnet v2的Faster RCNN
https://github.com/tensorflow/tensorflow/commit/055500bbcea60513c0160d213a10a7055f079312
mobil net https://github.com/tensorflow/models/tree/master/inception 准备数据 https://github.com/zehaos/MobileNet
https://github.com/balancap/SSD-Tensorflow
2017.9
https://github.com/udacity/CarND-Term1-Starter-Kit 环境配置
http://blog.csdn.net/xukai871105/article/details/39255089 树莓派mqtt
https://github.com/udacity/CarND-LaneLines-P1/blob/master/P1.ipynb
交通识别问题,行人识别，目标追踪
目标检测
- 传统方法 a.2001 paul viola 和Micahel jones 鲁棒实时目标检测 的 viola-jones 框架
b.梯度直方图 Hog
2.深度学习 a. overFeat 利用卷积 多尺度窗口滑动 b. r-cnn 选择性搜索 selective Search 提取可能目标；使用cnn 在该区域上提取特征；向量机分类 c. fast-rcnn 选择性搜索，cnn 提取特征， 区域兴趣池化 Region of interest ,ROI; 反向传播做分类和边框回归 d. yolo e. faster-rcnn cnn 提取特征;regio Proosal network 根据物体的分数来输出可能的目标；区域兴趣池化 Region of interest ,ROI pooling; 反向传播做分类和边框回归 f. SSd 在yolo 上改进，使用了多尺度特征图 g。 R-fcn 使用了 Faster-Rcnn的架构 https://tryolabs.com/blog/ 3.数据集 imageNet coco Pascal VOC Oxford-IIIT Pet kitti Vision
http://www.dev-c.com/nativedb/
github.com/osrf/car_demo
https://github.com/openai/roboschool
gym.openai.com
https://github.com/DartEnv/ddart-env
https://github.com/openai/baselines
目标跟踪
http://www.cs.cityu.edu.hk/~yibisong/iccv17/index.html convolutional Residual learning for visual tracking
pytorch caffe2 cntk 之间模型转换用onnx格式 github.com/onnx/onnx
github.com/nottombrown/rl-teacher https://github.com/nottombrown/rl-teacher.git
https://github.com/aleju/self-driving-truck
https://pan.baidu.com/s/1pL9J4Cz ros book
https://github.com/qboticslabs/ros_robotics_projects
https://cps-vo.org/group/CATVehicleTestbed/wiki
github.com/tigerneil/deep-reinforcement-learning-family
https://github.com/tigerneil/awesome-deep-rl
https://github.com/facebookresearch/ELF 开源游戏平台
https://github.com/tensorflow/models/blob/master/slim/nets/mobilenet_v1.md