2018 paper:https://github.com/amusi/daily-paper-computer-vision 人数统计：https://github.com/gjy3035/Awesome-Crowd-Counting pytorch中文文档：https://ptorch.com/news/53.html mac-shell 配置：https://www.jianshu.com/p/405956cdaca6 docker-安装cuda9-cudnn7：https://blog.csdn.net/u012939880/article/details/82887985 一文读懂BN：https://www.cnblogs.com/makefile/p/batch-norm.html?utm_source=debugrun&utm_medium=referral docker:https://blog.csdn.net/elaine_bao/article/details/53117676 BN层还可添加：https://blog.csdn.net/qq_29573053/article/details/79878437 roofline:https://zhuanlan.zhihu.com/p/34204282 BN层融合：https://www.cnblogs.com/nowgood/p/juan-ji-ceng-he-liang-hua-ceng-rong-he.html 读caffe源码过程：https://www.cnblogs.com/ycjing/p/5281384.html MPI多进程：https://blog.csdn.net/zouxy09/article/details/49031845#commentBox 标注工具汇总：https://blog.csdn.net/chaipp0607/article/details/79036312 108关键点人脸识别：https://github.com/zeusees/HyperLandmark vm配置：http://blog.csdn.net/u013142781/article/details/50529030 CMakeLists.txt：http://blog.csdn.net/z_h_s/article/details/50699905 莫烦python：https://morvanzhou.github.io/ 廖雪峰git与python：https://www.liaoxuefeng.com/ ～～ios上实现小规模yolo模型：http://colabug.com/107304.html 图像检测汇总：https://handong1587.github.io/deep_learning/2015/10/09/object-detection.html#t-cnn 压缩caffemodel：https://github.com/RalphMao/demo_nn_compress 实时车辆检测：https://github.com/upul/CarND-Vehicle-Detection MPII行人位姿数据集，可PK：http://human-pose.mpi-inf.mpg.de/#overview 识别领域汇总：http://www.cnblogs.com/zlslch/p/6970680.html xnor采访名言:https://techcrunch.com/2017/01/19/xnor-ai-frees-ai-from-the-prison-of-the-supercomputer/ 博客专栏：http://blog.csdn.net/column/mycolumn.html 多人位姿检测（骨骼点）2017：https://github.com/ZheC/Realtime_Multi-Person_Pose_Estimation github 57个最火热的框架：http://www.oschina.net/news/79500/57-most-popular-deep-learning-project-at-github 人脸检测、arm加速caffe：https://github.com/OAID/FaceRecognition 关于人脸检测的几个框架:http://www.sohu.com/a/154168717_206784 聊天机器人：https://www.cnblogs.com/zdz8207/p/DeepLearning-tensorflow-chatbot.html 大牛：http://www.shareditor.com/blogshow/?blogId=121 语料数据集 https://github.com/candlewill/Dialog_Corpus 各种数据集:https://zhuanlan.zhihu.com/p/25138563 车辆检测（RepNet）：https://github.com/xuqiantong/RepNet 反卷积：https://www.zhihu.com/question/43609045 条形码检测识别：https://github.com/seajune/Barcode-Scanner CVPR 2018：http://openaccess.thecvf.com/CVPR2018.py BN层讲解：https://blog.csdn.net/myarrow/article/details/51848285 sotfmax BP：https://www.cnblogs.com/alexanderkun/p/8098781.html inception v1-v4: http://baijiahao.baidu.com/s?id=1601882944953788623&wfr=spider&for=pc https://blog.csdn.net/wspba/article/details/68065564 mongo db:https://www.yiibai.com/mongodb/mongodb_python.html mqtt协议：https://www.jianshu.com/p/5c42cb0ed1e9
Boosted Tree：一篇很有见识的文章 http://dataunion.org/15787.html Kaggle—So Easy!百行代码实现排名Top 5%的图像分类比赛 ： http://blog.csdn.net/v_july_v/article/details/71598551 卷积的尺度和位置不变形解析：https://www.quora.com/How-is-a-convolutional-neural-network-able-to-learn-invariant-features
官方:http://colah.github.io/posts/2015-08-Understanding-LSTMs/ 翻译:http://www.jianshu.com/p/9dc9f41f0b29 cs231n老师的博客:http://karpathy.github.io/2015/05/21/rnn-effectiveness/ 这个是我见得写的最好的RNN，里面还有关于机器翻译，语音识别，语言模式，图像描述的论文，绝了：http://blog.csdn.net/heyongluoyao8/article/details/48636251 图像描述论文（cs231n的Andrej Karpathy）与github代码地址： http://cs.stanford.edu/people/karpathy/deepimagesent/ https://github.com/karpathy/neuraltalk
http://lib.csdn.net/article/deeplearning/57863?knId=1726 http://blog.csdn.net/samylee/article/details/53414108 V1版:http://blog.csdn.net/sinat_30071459/article/details/53100791
http://blog.csdn.net/sinat_30071459/article/details/51332084 http://blog.csdn.net/sinat_30071459/article/details/50723212
类似cs231n：http://blog.csdn.net/u014595019/article/details/52571966 大牛：http://blog.csdn.net/xbinworld/article/details/44464663 大牛！全：http://blog.csdn.net/hjimce/article/details/49255013 基础深度学习，有代码： https://www.zybuluo.com/hanbingtao/note/433855 Deep Learning Tutorials http://www.deeplearning.net/tutorial/ BN层原理：http://www.cnblogs.com/stingsl/p/6428694.html
学习指南翻译（版本1）： http://www.cnblogs.com/xueliangliu/archive/2013/04/03/2997437.html 学习指南翻译（版本2）： http://www.cnblogs.com/charleshuang/p/3648804.html 官方： http://deeplearning.net/software/theano/tutorial/ ★github示例： https://github.com/Newmu/Theano-Tutorials ★挺全的theano知识点讲解 http://www.cnblogs.com/shouhuxianjian/category/699462.html theano函数官方文档： http://deeplearning.net/software/theano/library/tensor/basic.html 简单易懂： http://www.cnblogs.com/YiXiaoZhou/p/6079428.html keras官方文档：http://keras-cn.readthedocs.io/en/latest/
super：http://www.cnblogs.com/lovemo1314/archive/2011/05/03/2035005.html format:http://www.jb51.net/article/63672.htm anaconda安装tensorflow： http://www.cnblogs.com/aloiswei/p/6510355.html
lua面向对象编程，写的很透彻： http://www.jellythink.com/archives/529 .与：区别： http://www.cnblogs.com/youxilua/archive/2011/07/28/2119059.html 元表方法： http://www.cnblogs.com/JimLy-BUG/p/5364281.html
https://github.com/dmlc/mxnet BMNet：https://github.com/hpi-xnor/BMXNet
卷积层源码：http://blog.csdn.net/shenxiaolu1984/article/details/52373174 tensor操作：http://blog.csdn.net/whgyxy/article/details/52204206 60分钟学torch： 原版：https://github.com/soumith/cvpr2015/blob/master/Deep Learning with Torch.ipynb 翻译：http://blog.csdn.net/baidu_17806763/article/details/61630538 torch学习资料：http://blog.csdn.net/victoriaw/article/details/71703568 torch学习系列（1），包含tensor操作：http://blog.csdn.net/whgyxy/article/category/6352333 torch学习系列（2）http://blog.csdn.net/u010946556/article/category/6216156 torch.nn模块学习（还有tensorflow、matlab、softmax问题）：http://blog.csdn.net/hejunqing14/article/category/6356970 torch学习系列（3）：http://blog.csdn.net/Hungryof/article/category/6245605 torch、自己建立一个层（如xnor）：https://zhuanlan.zhihu.com/p/21550685 官方：http://torch.ch/docs/developer-docs.html 一些torch小函数：http://blog.csdn.net/JIEJINQUANIL/article/category/5884857 关于torch建立层汇总： http://blog.csdn.net/lanran2/article/details/50494570 http://lib.csdn.net/article/deeplearning/51259 http://www.cnblogs.com/crossing/p/4826668.html http://torch.ch/docs/developer-docs.html
tensorflow与Keras（Keras中文官方文档）：https://keras-cn.readthedocs.io/en/latest/blog/keras_and_tensorflow/ Tensorflow-API：https://www.tensorflow.org/api_docs/ Faster-RCNN_TF:https://github.com/smallcorgi/Faster-RCNN_TF tf-faster-rcnn:https://github.com/endernewton/tf-faster-rcnn 将tensorflow移植到android手机实现物体识别！！http://blog.csdn.net/xiaopihaierletian/article/details/61933695 tensorflow-API：http://www.jianshu.com/nb/5517733 常用函数：http://blog.csdn.net/lenbow/article/details/52152766 tensorflow基础学习博客：http://blog.csdn.net/u012436149/article/details/53018924 tensorflow应用较全：http://blog.csdn.net/helei001/article/details/51842531
http://blog.csdn.net/wopawn/article/details/52223282 demo.py解析 论文解析，很好：http://blog.csdn.net/wopawn/article/details/52223282#reply 论文翻译，这人有一套rcnn博客：http://blog.csdn.net/u011534057/article/details/51259812 overfeat：http://blog.sciencenet.cn/blog-1583812-844178.html rpn代码解析：http://blog.csdn.net/sloanqin/article/details/51545125
http://blog.csdn.net/cryhelyxx/article/details/22514871
os.path:http://book.51cto.com/art/201405/440066.htm eastdict:https://pypi.python.org/pypi/easydict/ strip:http://www.cnblogs.com/itdyb/p/5046472.html zip:http://www.cnblogs.com/frydsh/archive/2012/07/10/2585370.html range与xrange:http://blog.csdn.net/ithomer/article/details/17285449 xml.etree.ElementTree:http://www.cnblogs.com/hongten/p/hongten_python_xml_etree_elementtree.html xml文件解析：http://blog.csdn.net/zhangjunbob/article/details/52769381 http://blog.csdn.net/gingerredjade/article/details/21944675 @作用：http://blog.sina.com.cn/s/blog_571b19a001013h7j.html stack:http://blog.csdn.net/huruzun/article/details/39801217 generate_anchors.py注释:http://blog.csdn.net/xzzppp/article/details/52317863 numpy.ravel() vs numpy.flatten() :http://blog.csdn.net/lanchunhui/article/details/50354978 meshgrid:http://blog.csdn.net/grey_csdn/article/details/69663432 anchor产生的问题：http://blog.csdn.net/zqjackking/article/details/59725989 bbox.pyx：http://blog.csdn.net/guotong1988/article/details/54729530 tf.nn.softmax_cross_entropy_with_logits：http://blog.csdn.net/mao_xiao_feng/article/details/53382790 tensor与ndarray操作与转换：http://blog.csdn.net/wyl1987527/article/details/62458057 ★变量、相关变量初始化、保存：http://www.cnblogs.com/claude-gyh/p/6554322.html saver：http://blog.csdn.net/u011500062/article/details/51728830 uninitial error:https://stackoverflow.com/questions/34001922/failedpreconditionerror-attempting-to-use-uninitialized-in-tensorflow ★与模型保存读取（saver）相关的一切操作：http://cv-tricks.com/tensorflow-tutorial/save-restore-tensorflow-models-quick-complete-tutorial/ 翻译：http://blog.163.com/wujiaxing009@126/blog/static/7198839920175671529472/ 其余相关：http://www.cnblogs.com/azheng333/archive/2017/06/09/6972619.html http://blog.csdn.net/thriving_fcl/article/details/71423039 http://www.jianshu.com/p/8487db911d9a http://blog.csdn.net/u014659656/article/details/53954793 tf.trainable_variables与tf.all_variables:http://blog.csdn.net/uestc_c2_403/article/details/72356448 tf.gradients 与 tf.stop_gradient() :http://blog.csdn.net/u012436149/article/details/53905797
利用OpenCV自带的haar training程序训练分类器 ：http://blog.csdn.net/carson2005/article/details/8171571 http://blog.csdn.net/liulina603/article/details/8184451 haar原理：http://www.cnblogs.com/ello/archive/2012/04/28/2475419.html http://blog.csdn.net/bbzz2/article/details/50764159 dlib+opencv人脸特征检测：http://blog.csdn.net/zmdsjtu/article/details/52422847 训练常见问题：http://www.cnblogs.com/chensheng-zhou/p/5542887.html
https://stackoverflow.com/questions/29118317/how-to-convert-mat-to-array2drgb-pixel http://blog.csdn.net/huixingshao/article/details/55510950 http://blog.csdn.net/qq_26671711/article/details/53713529 http://blog.csdn.net/yubin1277408629/article/details/53561037
vcglib -github：https://github.com/cnr-isti-vclab/vcglib win10+vs+cuda ：http://blog.csdn.net/u011821462/article/details/50145221 ★三维可视化工程：http://redwood-data.org/indoor/tutorial.html win10+vs2015 meshlab编译 ：http://blog.csdn.net/hanshuobest/article/details/71525388 windows-git：http://blog.csdn.net/qq_34698126/article/details/53521187
kinfu安装配置全解：http://blog.csdn.net/alan_1550587588/article/details/54582192 VS2010和IVF2011的安装教程 Fortran： http://wenku.baidu.com/view/9e28de1b01f69e314332949c.html Fortran认证书Intel_Visual_Fortran_XE2011.lic： http://pan.baidu.com/s/1kT9lR8r ros下使用kinfu：http://blog.csdn.net/l_h2010/article/details/38349927
c++笔记：http://www.cnblogs.com/ggjucheng/archive/2012/08/18/2645319.html typedef：http://blog.csdn.net/ameyume/article/details/6326278 指针的指针：http://blog.jobbole.com/60647/ 成员函数指针：http://blog.csdn.net/jinjinclouded/article/details/5189540 c++指针：http://www.cnblogs.com/ggjucheng/archive/2011/12/13/2286391.html qt学习：http://www.kuqin.com/qtdocument/tutorial.html argc和argv含义及用法 ：http://blog.csdn.net/dcrmg/article/details/51987413 虚函数与纯虚函数：http://blog.csdn.net/xwpc702/article/details/8670025 http://blog.csdn.net/hackbuteer1/article/details/7558868
安装中出现问题解决（中文）博客：https://www.codelast.com/原创-在ubuntu-14-04系统中为ell安装python-3-6-通过miniconda/ https://www.codelast.com/原创-在树莓派3上使用微软ell嵌入式学习库1/
http://blog.csdn.net/chentravelling/article/details/50924144 http://blog.csdn.net/taily_duan/article/details/51011288 http://blog.csdn.net/zouxy09/article/details/8683859 http://blog.csdn.net/dcrmg/article/details/52684477 http://blog.csdn.net/crzy_sparrow/article/details/7407604 加载深度信息求速度原理：
【人体姿态】Stacked Hourglass算法详解 （torch代码）：http://blog.csdn.net/shenxiaolu1984/article/details/51428392 Object Detection︱RCNN、faster-RCNN框架的浅读与延伸内容笔记 ： http://blog.csdn.net/roslei/article/details/73459873 ipython-notebook caffe识别与可视化：http://nbviewer.jupyter.org/github/BVLC/caffe/blob/master/examples/00-classification.ipynb#Classification:-Instant-Recognition-with-Caffe
FCN-github：https://github.com/shelhamer/fcn.berkeleyvision.org fcn-知乎：https://zhuanlan.zhihu.com/p/22976342 FCN-CRF：https://zhuanlan.zhihu.com/p/22308032
SqueezeNet-github:https://github.com/DeepScale/SqueezeNet Deep Compression阅读理解及Caffe源码修改 ：http://blog.csdn.net/may0324/article/details/52935869 阅读理解： http://blog.csdn.net/zijin0802034/article/details/53982812 http://blog.csdn.net/boon_228/article/details/51718521 http://blog.csdn.net/cyh_24/article/details/51708469 http://blog.csdn.net/joshuaxx316/article/details/52514978 压缩caffemodel：https://github.com/RalphMao/demo_nn_compress SqueezeNet+Faster-RCNN+OHEM：http://blog.csdn.net/u011956147/article/details/53714616
https://github.com/CharlesShang/TFFRCNN，内含PVANET模型 运行TFFRCNN：http://blog.csdn.net/yuexin2/article/details/75019330 TFRRCNN训练自己的模型：http://blog.csdn.net/yuexin2/article/details/75025720
CNN 模型压缩与加速算法综述 :http://blog.csdn.net/QcloudCommunity/article/details/77719498 YOLO对比SVM+HOG 耗时更短 处理更精确（附github资源）：http://www.weiot.net/article-59672.html LCDet：http://blog.csdn.net/zhangjunhit/article/details/72831440 PVANet（TFFRCNN中有）：https://arxiv.org/pdf/1608.08021.pdf http://blog.csdn.net/wydbyxr/article/details/68931337 mobilenet：http://blog.csdn.net/zchang81/article/details/73321202 //（有实现链接）http://blog.csdn.net/mao_feng/article/details/75116085 //论文翻译：https://baijiahao.baidu.com/s?id=1566004753349359&wfr=spider&for=pc ARM-NEON：http://blog.csdn.net/qiek/article/details/50900890 http://blog.csdn.net/ljp12345/article/details/53490094 popcnt：http://www.cnblogs.com/zyl910/archive/2012/11/02/testpopcnt.html
https://www.zhihu.com/question/39921464 https://www.zhihu.com/question/53697001 https://www.zhihu.com/question/27872849
http://blog.csdn.net/mao_kun/article/details/54882528 http://blog.csdn.net/u010620604/article/details/52464529
http://www.360doc.com/content/12/0507/10/9369336_209205930.shtml http://blog.csdn.net/zhubaohua_bupt/article/details/52760411
david 9:http://nooverfit.com/wp/15-增强学习101-闪电入门-reinforcement-learning/ 精彩:http://www.cnblogs.com/jinxulin/p/3517377.html 知乎大神:https://www.zhihu.com/question/41775291
https://zhuanlan.zhihu.com/p/24816781
https://github.com/allenai/XNOR-Net/issues https://github.com/jiecaoyu/XNOR-Net-PyTorch https://github.com/tensorflow/tensorflow/issues/1592 https://github.com/MatthieuCourbariaux/BinaryNet https://github.com/loswensiana/BWN-XNOR-caffe/tree/master/examples/imagenet https://github.com/tensorflow/tensorflow/tree/master/third_party/eigen3/unsupported/Eigen/CXX11/src/FixedPoint ☆修改梯度相关 http://blog.csdn.net/buyi_shizi/article/details/51512848 http://blog.csdn.net/yihaizhiyan/article/details/44159063
详细版本：http://blog.csdn.net/jiongnima/article/details/78549326?locationNum=1&fps=1 http://blog.csdn.net/zoro_lov3/article/details/74550735 http://blog.csdn.net/z13653662052/article/details/70949440 http://blog.csdn.net/supe_king/article/details/58121993 http://blog.csdn.net/u010402786/article/details/72883421 http://www.cnblogs.com/xuanxufeng/p/6243342.html https://github.com/315386775/FCN_train https://github.com/msracver/FCIS
http://blog.csdn.net/totodum/article/details/51059380 http://blog.csdn.net/fishman_yinwang/article/details/78029309 http://blog.csdn.net/jiaojialulu/article/details/77430563
https://github.com/jfkirk/tensorrec https://github.com/bnak/Recommendations_Engine https://github.com/songgc/TF-recomm https://github.com/sonyisme/keras-recommendation https://github.com/geeky-bit/Recommendation_systems-using-SVD-KNN-etc https://github.com/amzn/amazon-dsstne recommendation decomposition
https://www.douban.com/note/510047571/ https://github.com/Lockvictor/MovieLens-RecSys
http://blog.csdn.net/u011467621/article/details/48624973 https://github.com/joeyqzhou/recommendation-system
讲解：http://blog.csdn.net/joycewyj/article/details/51692976
https://github.com/chengstone/movie_recommender http://blog.csdn.net/chengcheng1394/article/details/78820529
https://github.com/loswensiana/BWN-XNOR-caffe 添加新层： https://blog.csdn.net/shuzfan/article/details/51322976（cutoff） https://blog.csdn.net/wfei101/article/details/76735760（盗版） https://blog.csdn.net/kuaitoukid/article/details/41865803（maxout与NIN） https://blog.csdn.net/xizero00/article/details/52529341（图像缩放） https://blog.csdn.net/happyflyy/article/details/54866037（maxout） caffe函数解析： https://blog.csdn.net/seven_first/article/details/47378697
caffe卷积实现原理： https://blog.csdn.net/jiongnima/article/details/69055941 im2col原理： https://blog.csdn.net/jiongnima/article/details/69736844
https://github.com/ternaus/TernausNet https://github.com/ternaus/robot-surgery-segmentation
很多篇都有用：https://blog.csdn.net/tangwei2014?viewmode=contents 添层（facenet）：https://blog.csdn.net/langb2014/article/details/50489305 https://blog.csdn.net/seven_first/article/details/50272005 caffe源码解析：https://blog.csdn.net/seven_first/article/category/5721883
https://github.com/BoyuanJiang/Age-Gender-Estimate-TF
https://blog.csdn.net/jiongnima/article/category/7245212
https://github.com/fateleak/awesome-chatbot-list
github: https://github.com/dhernandez0/sgm https://github.com/fixstars/libSGM https://github.com/WanchaoYao/SGM 博客： https://blog.csdn.net/zilanpotou182/article/details/73382412 https://blog.csdn.net/wsj998689aa/article/details/49464017 https://www.cnblogs.com/hrlnw/p/4746170.html https://blog.csdn.net/qq_31785865/article/details/78451880 https://blog.csdn.net/laobai1015/article/details/52527233 http://lib.csdn.net/article/datastructure/10208
cblas_sgemm：https://blog.csdn.net/ZhikangFu/article/details/78258393 conv层讲解：https://blog.csdn.net/sinat_22336563/article/details/69808612 caffe讲解全（1）：https://blog.csdn.net/seven_first/article/details/47378697#1-caffecpugemm-函数 caffe讲解全（2）：https://blog.csdn.net/langb2014/article/details/51558515
caffe讲解全：https://blog.csdn.net/jiongnima/article/category/6436731
pyrorch：https://github.com/jiecaoyu/XNOR-Net-PyTorch/tree/7f84f1bf6921bc99b05bbc869dddf2bd64989a40 tensorflow：https://github.com/AngusG/tensorflow-xnor-bnn 问题：https://github.com/jiecaoyu/XNOR-Net-PyTorch/issues/13 反向传播：https://github.com/jiecaoyu/XNOR-Net-PyTorch/blob/master/notes/notes.pdf
boost::thread https://www.jianshu.com/p/0b2e360243f4 https://www.jianshu.com/p/30e79d2a8f7d https://blog.csdn.net/zhuxiaoyang2000/article/details/6588031/
pthread http://www.runoob.com/cplusplus/cpp-multithreading.html https://blog.csdn.net/hudashi/article/details/7709413 http://www.docin.com/p-65547314.html https://www.ibm.com/developerworks/cn/linux/l-pthred/index.html http://www.cppblog.com/saha/articles/189802.html
https://download.csdn.net/download/jieweijason/10200330 https://developer.arm.com/technologies/neon/intrinsics https://blog.csdn.net/may0324/article/details/72847800 https://www.jianshu.com/p/68879baa7c1f?from=timeline&isappinstalled=0
https://www.cnblogs.com/hrlnw/p/3767853.html https://www.cnblogs.com/hrlnw/p/3723072.html
prototxt： https://github.com/suzhenghang/MobileNetv2/tree/master/.gitignore https://github.com/austingg/MobileNet-v2-caffe mobilenet-sdd： https://github.com/PINTO0309/MobileNet-SSD https://github.com/PINTO0309/MobileNet-SSDLite-RealSense-TF
https://github.com/eric612/MobileNet-YOLO https://github.com/PINTO0309/MobileNet-SSD https://github.com/chuanqi305/MobileNet-SSD https://github.com/chuanqi305/MobileNetv2-SSDLite https://github.com/FreeApe/VGG-or-MobileNet-SSD https://github.com/farmingyard/ShuffleNet https://github.com/linchaozhang/shufflenet-ssd
https://github.com/chuanqi305/SqueezeNet-SSD https://github.com/ChenYingpeng/caffe-ssd-frameworks
https://www.ctolib.com/ZhishengWang-Embedded-Neural-Network.html
1、多进程：https://blog.csdn.net/zouxy09/article/details/49031845#commentBox 2、gamma矫正：https://blog.csdn.net/linqianbi/article/details/78617615 3、仿射变化：https://blog.csdn.net/wsx_9999/article/details/80441125
dw_conv比conv慢：https://blog.csdn.net/liu1152239/article/details/81478463 conv bn融合：https://www.cnblogs.com/nowgood/p/juan-ji-ceng-he-liang-hua-ceng-rong-he.html
https://blog.csdn.net/Pattorio/article/details/80095511 https://blog.csdn.net/m_buddy/article/details/82926024 https://github.com/Jumabek/darknet_scripts
dbb转voc https://github.com/leeesangwon/bdd100k_to_VOC https://blog.csdn.net/qq_17278169/article/details/82189776 voc转lmdb https://blog.csdn.net/jiang_ming_/article/details/81068854 https://blog.csdn.net/LOVExinxinsensen/article/details/80982423 txt转voc https://blog.csdn.net/pursuit_zhangyu/article/details/79949947 https://blog.csdn.net/yjl9122/article/details/56842098
https://spaces.ac.cn/archives/5693 https://www.jianshu.com/p/003c5b6477ba https://blog.csdn.net/stezio/article/details/87816785
https://blog.csdn.net/xiaozhu_1024/article/details/80585151 https://www.linuxidc.com/Linux/2016-01/127857.htm
centernet pytorch1.0+版本问题： https://www.gitmemory.com/issue/xingyizhou/CenterNet/7/486653333
