姿勢估計實驗-Realtime_Multi-Person_Pose_Estimation-CMU

前言：

論文及源代碼網址：

https://github.com/ZheC/Realtime_Multi-Person_Pose_Estimation

 

地址2:

https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/doc/installation.md

 

---

 

1.  一個失敗的嘗試

首先是跑一個例子試試，使用

https://github.com/CMU-Perceptual-Computing-Lab/openpose/

 

按照作者的指導，如下：

Training Steps

• Run cd training; bash getData.sh to obtain the COCO images in dataset/COCO/images/, keypoints annotations in dataset/COCO/annotations/ and COCO official toolbox in dataset/COCO/coco/.
  • 　　這里建議開screen，因為要下載超級久
• Run getANNO.m in matlab to convert the annotation format from json to mat in dataset/COCO/mat/.
• Run genCOCOMask.m in matlab to obatin the mask images for unlabeled person. You can use 'parfor' in matlab to speed up the code.
  • 　　注意修改Matlab程序中的路徑，在  im = imread(['dataset/COCO/', img_paths]);的COCO后面加上images/，原程序有點問題
  •       這里嘗試開啟了一下並行運算，速度特別快，將代碼中第二個for循環改為parfor i = 1:L，然后使用Matlab的時候，先輸入matlabpool open，然后再運行程序。關閉的方法為matlabpool close
    • 　　然后遇到了Subscripted assignment between dissimilar structures.問題，看代碼也沒有找到問題，參考 [1]
• Run genJSON('COCO') to generate a json file in dataset/COCO/json/ folder. The json files contain raw informations needed for training.
• Run python genLMDB.py to generate your LMDB. (You can also download our LMDB for the COCO dataset (189GB file) by: bash get_lmdb.sh)
• Download our modified caffe: caffe_train. Compile pycaffe. It will be merged with caffe_rtpose (for testing) soon.
• Run python setLayers.py --exp 1 to generate the prototxt and shell file for training.
• Download VGG-19 model, we use it to initialize the first 10 layers for training.
• Run bash train_pose.sh 0,1 (generated by setLayers.py) to start the training with two gpus.

---

 

2. 安裝Caffe

這種跑一個例子失敗之后，

轉而使用最根本的方法從頭按照步驟編譯，不走捷徑：參考下面的網址

https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/doc/installation.md

因為電腦里已經安裝過caffe的依賴環境了，所以把caffe編譯一下就可以用了，參考下面的話 [3] 

### Install Caffe ###
cd 3rdparty/caffe/
# Select your desired Makefile file (run only one of the next 4 commands)
cp Makefile.config.Ubuntu14_cuda7.example Makefile.config # Ubuntu 14, cuda 7
cp Makefile.config.Ubuntu14_cuda8.example Makefile.config # Ubuntu 14, cuda 8
cp Makefile.config.Ubuntu16_cuda7.example Makefile.config # Ubuntu 16, cuda 7
cp Makefile.config.Ubuntu16_cuda8.example Makefile.config # Ubuntu 16, cuda 8
# Change any custom flag from the resulting Makefile.config (e.g. OpenCV 3, Atlas/OpenBLAS/MKL, etc.)
# Compile Caffe

make all -j${number_of_cpus} && make distribute -j${number_of_cpus}

比如make all -j16（使用16個CPU同時運行）



 3. 安裝 OpenPose
cd ../../models/
bash ./getModels.sh # It just downloads the Caffe trained models
cd ..
# Same file cp command as the one used for Caffe
cp ubuntu/Makefile.config.Ubuntu14_cuda7.example Makefile.config
# Change any custom flag from the resulting Makefile.config (e.g. OpenCV 3, Atlas/OpenBLAS/MKL, etc.)
make all -j${number_of_cpus}

一定要嚴格按照順序（實驗室服務器用的是zsh，所以注意替換bash為zsh）

 

如果使用OpenCV 3版本的話，可以通過所有make，但是不能運行實際的例子

按照作者在某個Issue下的回答，應該使用OpenCV2.4，然后折騰很久，見另一篇博客 [7]

然后出現無法make，錯誤中關鍵行為這個

../lib/libcaffe.so: undefined reference to cv::imread(cv::String const&, int)'

搜索之后得到解答：[4]

open your Makefile with some text editor, locate line 164 (in my case), add opencv_imgcodecs behind.

 LIBRARIES += glog gflags protobuf leveldb snappy \
  lmdb boost_system hdf5_hl hdf5 m \
  opencv_core opencv_highgui opencv_imgproc opencv_imgcodecs

在vim里使用

:se nu

然后回車，可以顯示行號；使用

/opencv

可以快速定位到第一個出現的opencv字符串的位置

 

添加后可以make了，然后不出意料的還是不能運行實際的例子：

 

4. 運行一個例子

實際例子運行方法：[3]

1. Running on Video

# Ubuntu
./build/examples/openpose/openpose.bin --video examples/media/video.avi
# With face and hands
./build/examples/openpose/openpose.bin --video examples/media/video.avi --face --hand

會出現

zsh: abort (core dumped)  ./build/examples/openpose/openpose.bin --video examples/media/video.avi

再試試別的運行方法：

3. Running on Images

# Ubuntu
./build/examples/openpose/openpose.bin --image_dir examples/media/
# With face and hands
./build/examples/openpose/openpose.bin --image_dir examples/media/ --face --hand

會出現

Starting pose estimation demo.

 

Error:

boost::filesystem::directory_iterator::construct: No such file or directory

 

Coming from:

- src/openpose/utilities/fileSystem.cpp:getFilesOnDirectory():186

- src/openpose/producer/imageDirectoryReader.cpp:getImagePathsOnDirectory():25

- src/openpose/utilities/flagsToOpenPose.cpp:flagsToProducer():133

terminate called after throwing an instance of 'std::runtime_error'

  what():  

Error:

boost::filesystem::directory_iterator::construct: No such file or directory

 

Coming from:

- src/openpose/utilities/fileSystem.cpp:getFilesOnDirectory():186

- src/openpose/producer/imageDirectoryReader.cpp:getImagePathsOnDirectory():25

- src/openpose/utilities/flagsToOpenPose.cpp:flagsToProducer():133

 

zsh: abort (core dumped)  ./build/examples/openpose/openpose.bin --image_dir examples/media/

然后搜索后，根據 [5] DuinoDu的回答，去 [6] 里找答案，“install boost from src.”

根據user3715812的回答，輸入以下命令行來安裝BOOST：

wget -O boost_1_55_0.tar.gz http://sourceforge.net/projects/boost/files/boost/1.55.0/boost_1_55_0.tar.gz/download
tar xzvf boost_1_55_0.tar.gz
cd boost_1_55_0/

./bootstrap.sh --prefix=/usr/local     // 這里改成自己想要安裝到的目錄
./b2
./b2 install       // 沒有SUDO權限但是也可以

然后再去運行

./build/examples/openpose/openpose.bin --image_dir examples/media/

這次的報錯和上次不一樣了，是

Starting pose estimation demo.
Auto-detecting GPUs... Detected 16 GPU(s), using them all.
Starting thread(s)
: cannot connect to X server 
E0819 00:20:53.031266 20765 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20766 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20771 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20773 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20767 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20775 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20776 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20777 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20778 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20765 common.cpp:121] Cannot create Curand generator. Curand won't be available.
E0819 00:20:53.031266 20769 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20768 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20774 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20770 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20779 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20772 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20779 common.cpp:121] Cannot create Curand generator. Curand won't be available.
E0819 00:20:53.031266 20771 common.cpp:121] Cannot create Curand generator. Curand won't be available.
E0819 00:20:53.031266 20773 common.cpp:121] Cannot create Curand generator. Curand won't be available.
F0819 00:20:53.031266 20779 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.031266 20771 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.035267 20773 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting down
*** Check failure stack trace: ***
E0819 00:20:53.031266 20780 common.cpp:114] Cannot create Cublas handle. Cublas won't be available.
E0819 00:20:53.031266 20778 common.cpp:121] Cannot create Curand generator. Curand won't be available.
E0819 00:20:53.031266 20775 common.cpp:121] Cannot create Curand generator. Curand won't be available.
E0819 00:20:53.031266 20767 common.cpp:121] Cannot create Curand generator. Curand won't be available.
F0819 00:20:53.031266 20779 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.031266 20771 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.035267 20773 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.035267 20778 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.035267 20775 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting downF0819 00:20:53.035267 20767 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting down
*** Check failure stack trace: ***
F0819 00:20:53.031266 20765 common.cpp:152] Check failed: error == cudaSuccess (29 vs. 0)  driver shutting down
zsh: abort (core dumped)  ./build/examples/openpose/openpose.bin --image_dir examples/media/

 

然后在GITHUB 的ISSUE區准備提問的時候，根據提示

Note: add --logging_level 0 to get higher debug information.

 在運行的命令行后加上了這個，即：

./build/examples/openpose/openpose.bin --image_dir examples/media/  --logging_level 0

於是得到了較為准確的信息：

可以理解為又一個圖形界面要展示但是出不來 [8]（X server相關), 

查閱資料后根據 [9] 的xiaoyan的回答：

Add "export DISPLAY=your ip:0.0" in ~/.bashrc, source ~/.bashrc (ip is your host IP, where you wanna display plots or things like that. If you are using Windows, cmd-ipconfig, find your IP).

 於是不再出現X server問題了，出現了下圖的問題

 

找了好久沒找到問題，最后求助了組里的史博士，這里表示非常感謝！！！

原因是這個Demo默認使用所有探測到的CUDA 設備，而實驗室的CUDA設備還有很多人在用，這會造成問題（可能是沖突或者資源不夠，或者不被允許個人使用這么多個？）

所以在運行的命令行前面要加上一定的限制：

CUDA_VISIBLE_DEVICES=2 ./build/examples/openpose/openpose.bin --net_resolution "160x80" --video examples/media/video.avi

使得被探測到的設備數量只有兩個。

然后發現還是運行不了，為什么呢？

因為OS X上的SHELL並不具備顯示遠程窗口的功能，

所以在史博士的推薦下使用了MobaXterm（在WIN10下），然后就成功了，因為這個軟件直接內置了X server的接口

 

更新：如果想在Mac上運行，可以參考

[10] http://blog.csdn.net/dobell/article/details/55047811

 

安裝XQuartz，
https://www.xquartz.org
按說明安裝好

打開 mac terminal 
ssh -X   {用戶名}@{遠程端ip}； （注意大寫的X）
這次就可以了

 

 

實際運行截圖如下： 

 

你也可以讀一下我在GITHUB上的提問：
https://github.com/CMU-Perceptual-Computing-Lab/openpose/issues/184
可能會有幫助

---

 

另外：

因為實驗室的服務器出於安全考慮不允許使用

sudo pip install

, 所以使用

pip install --user

來替代。[2]

  

參考文獻：

[1] http://www.cnblogs.com/emituofo/archive/2011/11/13/2247523.html

[2] https://stackoverflow.com/questions/31512422/pip-install-r-oserror-errno-13-permission-denied

[3] https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/master/doc/installation.md

[4] https://github.com/BVLC/caffe/issues/1276

[5] https://github.com/CMU-Perceptual-Computing-Lab/openpose/issues/17

[6] https://stackoverflow.com/questions/12578499/how-to-install-boost-on-ubuntu

[7] http://www.cnblogs.com/QingHuan/p/7278357.html

[8] http://www.cnblogs.com/super119/archive/2010/12/18/1910065.html

[9] https://askubuntu.com/questions/571116/cannot-connect-to-x-server-error

[10] http://blog.csdn.net/dobell/article/details/55047811