Xiaoxu Meng – Welcome to blog of Xiaoxu Meng :) Please visit http://www.mengxiaoxu.com for more information. Thanks!

Install OpenEXR for Python (Windows and Ubuntu)

Windows

Download the wheel for openEXR from:

Python Extension Packages for Windows – Christoph Gohlke (uci.edu)

python -m pip install OpenEXR-1.3.0-cp36-cp36m-win_amd64.whl

1 2	python -m pip install OpenEXR-1.3.0-cp36-cp36m-win_amd64.whl

Ubuntu:

sudo apt-get install zlib1g-dev
sudo apt-get install libopenexr-dev
sudo apt-get install openexr

sudo apt-get install zlib1g-dev

sudo apt-get install libopenexr-dev

sudo apt-get install openexr

How to print the gradient of intermediate variables in Pytorch

grads = {}
def save_grad(name):
    def hook(grad):
        grads[name] = grad
    return hook

x = Variable(torch.randn(1,1), requires_grad=True)
y = 3*x
z = y**2

# In here, save_grad('y') returns a hook (a function) that keeps 'y' as name
y.register_hook(save_grad('y'))
z.register_hook(save_grad('z'))
z.backward()

print(grads['y'])
print(grads['z'])

grads = {}

def save_grad(name):

def hook(grad):

grads[name] = grad

return hook

x = Variable(torch.randn(1,1), requires_grad=True)

y = 3*x

z = y**2

# In here, save_grad('y') returns a hook (a function) that keeps 'y' as name

y.register_hook(save_grad('y'))

z.register_hook(save_grad('z'))

z.backward()

print(grads['y'])

print(grads['z'])

Thanks to Adam Paszke’s post in Pytorch Discussion

I struggled with a problem today: My parameter “b” is not updating in the following code:

b = nn.Parameter(torch.ones(batch_size, 1)
a = torch.norm(b, dim=1)

1 2	b = nn.Parameter(torch.ones(batch_size, 1) a = torch.norm(b, dim=1)

There’s nothing wrong with the gradient of “a”. So what’s the problem?

The problem is: I used the wrong initialization of “b”. I init “b” with all zeros. and the gradient of the norm of an all-zero vector is always zero.

Error about multiply with different precision

Today I met an error:

a is a float32 pytorch tensor; and b is a float64 pytorch tensor with all elements equals 1. When I perform the following multiplication:

c = a * b

c = a * b

The result of “c” is 0.

The error will be fixed when b is converted to float32.

DEBUG: unable to execute ‘:/usr/local/cuda/bin/nvcc’: No such file or directory error: command ‘:/usr/local/cuda/bin/nvcc’ failed with exit status 1

zWhen trying to install SoftRas & neural renderer, I got error:

unable to execute ‘:/usr/local/cuda/bin/nvcc’: No such file or directory

error: command ‘:/usr/local/cuda/bin/nvcc’ failed with exit status 1

Solution:

-export CUDA_HOME=$CUDA_HOME:/usr/local/cuda 
+export CUDA_HOME=/usr/local/cuda

1 2	-export CUDA_HOME=$CUDA_HOME:/usr/local/cuda +export CUDA_HOME=/usr/local/cuda

Install openpose on ubuntu 16.04

Config:
System: Ubuntu 16.04
CUDA: 10.0
Graphic Crad: RTX 2080
——————————————————————————————————————————

Download openpose and dependencies:

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

1

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

If “–recursive” is not added here, the default caffe will not be downloaded!
cmake .. make all make install

1
2
3

cmake ..
make all
make install

The process will not be so smooth!!

Error:

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.
Please set them or make sure they are set and tested correctly in the CMake files:
CUDA_curand_LIBRARY (ADVANCED)
    linked by target "caffe" in directory /home/xiaoxu/Documents/rgb2mesh/Borrowed/openpose/3rdparty/caffe/src/caffe

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.

Please set them or make sure they are set and tested correctly in the CMake files:

CUDA_curand_LIBRARY (ADVANCED)

linked by target "caffe" in directory /home/xiaoxu/Documents/rgb2mesh/Borrowed/openpose/3rdparty/caffe/src/caffe

Solve: open “openpose/build/caffe/src/openpose_lib-build/CMakeCache.txt” with cmake-gui
Change cuda-9.0 to cuda-10.0

Error:

(pytorch_py3) xiaoxu@chuan:~/Documents/rgb2mesh/Borrowed/openpose/build/examples/tutorial_api_python$ python 01_body_from_image.py 
Error: OpenPose library could not be found. Did you enable `BUILD_PYTHON` in CMake and have this Python script in the right folder?
Traceback (most recent call last):
  File "01_body_from_image.py", line 26, in 
    raise e
  File "01_body_from_image.py", line 23, in 
    from openpose import pyopenpose as op
  File "../../python/openpose/__init__.py", line 1, in 
    from . import pyopenpose as pyopenpose
ImportError: cannot import name 'pyopenpose'

(pytorch_py3) xiaoxu@chuan:~/Documents/rgb2mesh/Borrowed/openpose/build/examples/tutorial_api_python$ python 01_body_from_image.py

Error: OpenPose library could not be found. Did you enable `BUILD_PYTHON` in CMake and have this Python script in the right folder?

Traceback (most recent call last):

File "01_body_from_image.py", line 26, in

raise e

File "01_body_from_image.py", line 23, in

from openpose import pyopenpose as op

File "../../python/openpose/__init__.py", line 1, in

from . import pyopenpose as pyopenpose

ImportError: cannot import name 'pyopenpose'

Solve: Goto “build/python/openpose” and run “make”

(pytorch_py3) xiaoxu@chuan:~/Documents/rgb2mesh/Borrowed/openpose/build/python/openpose$ make
[ 97%] Built target openpose
[100%] Built target pyopenpose

(pytorch_py3) xiaoxu@chuan:~/Documents/rgb2mesh/Borrowed/openpose/build/python/openpose$ make

[ 97%] Built target openpose

[100%] Built target pyopenpose

Install Caffe

Configuration:
Ubuntu 16.04
CUDA 10.0
GTX 2080

The reference is the step-by-step tutorial.
I met one error:

To solve it, two steps are necessary:

https://blog.csdn.net/fdd096030079/article/details/84451811
https://stackoverflow.com/questions/48383846/nvcc-fatal-unsupported-gpu-architecture-compute-20-while-cuda-9-1caffeopen

Connect to wifi using commend line

You know, the ubuntu system in my Great Alienware is not healthy. Many functions are not working well, including the wifi connection. It is impossible to connect to a new wifi using the graphical interface. I will get error like:

Image result for active connection removed before it was initialized

I solved the connection issue using

nmcil

nmcil

Reference

Determine the name of the wifi. In many tutorials, they directly call their wifi “wlan0”. However, the name is different on my machine. Run the following command:

(base) xiaoxu@chuan:~$ nmcli d
DEVICE  TYPE      STATE         CONNECTION 
wlp4s0  wifi      disconnected  --         
enp5s0  ethernet  unavailable   --         
lo      loopback  unmanaged     --

(base) xiaoxu@chuan:~$ nmcli d

DEVICE TYPE STATE CONNECTION

wlp4s0 wifi disconnected --

enp5s0 ethernet unavailable --

lo loopback unmanaged --

The name of my wifi is “wlp4s0”.

List the wifi networks

(base) xiaoxu@chuan:~$ nmcli d wifi list
*  SSID                   MODE   CHAN  RATE       SIGNAL  BARS  SECURITY         
   xfinitywifi            Infra  6     54 Mbit/s  84      ▂▄▆█                   
   HOME-CBE2              Infra  6     54 Mbit/s  82      ▂▄▆█  WPA1 WPA2        
   Neptune_EXT            Infra  1     54 Mbit/s  77      ▂▄▆_  WPA1 WPA2        
   NETGEAR91              Infra  11    54 Mbit/s  60      ▂▄▆_  WPA2             
   Fios-TVB5U             Infra  11    54 Mbit/s  54      ▂▄__  WPA2             
   2NXG7                  Infra  6     54 Mbit/s  45      ▂▄__  WPA2             
   Zeus_EXT               Infra  153   54 Mbit/s  45      ▂▄__  WPA1 WPA2        
   FiOS-A5O2K             Infra  6     54 Mbit/s  40      ▂▄__  WPA2             
   nevetica               Infra  6     54 Mbit/s  40      ▂▄__  WPA1 WPA2        
   --                     Infra  6     54 Mbit/s  39      ▂▄__  WPA1 WPA2        
   Fios-TVB5U-5G          Infra  161   54 Mbit/s  35      ▂▄__  WPA2             
   --                     Infra  161   54 Mbit/s  35      ▂▄__  WPA2             
   --                     Infra  1     54 Mbit/s  30      ▂___  WPA1 WPA2        
   DIRECT-PG-FireTV_3568  Infra  153   54 Mbit/s  30      ▂___  WPA2             
   --                     Infra  1     54 Mbit/s  29      ▂___  WPA1 WPA2 802.1X 
   --                     Infra  1     54 Mbit/s  27      ▂___                   
   OutOfService           Infra  11    54 Mbit/s  27      ▂___  WPA2 802.1X      
   --                     Infra  161   54 Mbit/s  27      ▂___  WPA2             
   Axify                  Infra  11    54 Mbit/s  25      ▂___  WPA2             
   sergek5                Infra  36    54 Mbit/s  25      ▂___  WPA1 WPA2        
   xfinitywifi            Infra  36    54 Mbit/s  24      ▂___                   
   XFINITY                Infra  36    54 Mbit/s  24      ▂___  WPA2 802.1X      
   --                     Infra  36    54 Mbit/s  24      ▂___  WPA1 WPA2        
   FiOS-A5O2K-5G          Infra  161   54 Mbit/s  24      ▂___  WPA2             
   --                     Infra  36    54 Mbit/s  22      ▂___  WPA1 WPA2 802.1X

(base) xiaoxu@chuan:~$ nmcli d wifi list

* SSID MODE CHAN RATE SIGNAL BARS SECURITY

xfinitywifi Infra 6 54 Mbit/s 84 ▂▄▆█

HOME-CBE2 Infra 6 54 Mbit/s 82 ▂▄▆█ WPA1 WPA2

Neptune_EXT Infra 1 54 Mbit/s 77 ▂▄▆_ WPA1 WPA2

NETGEAR91 Infra 11 54 Mbit/s 60 ▂▄▆_ WPA2

Fios-TVB5U Infra 11 54 Mbit/s 54 ▂▄__ WPA2

2NXG7 Infra 6 54 Mbit/s 45 ▂▄__ WPA2

Zeus_EXT Infra 153 54 Mbit/s 45 ▂▄__ WPA1 WPA2

FiOS-A5O2K Infra 6 54 Mbit/s 40 ▂▄__ WPA2

nevetica Infra 6 54 Mbit/s 40 ▂▄__ WPA1 WPA2

-- Infra 6 54 Mbit/s 39 ▂▄__ WPA1 WPA2

Fios-TVB5U-5G Infra 161 54 Mbit/s 35 ▂▄__ WPA2

-- Infra 161 54 Mbit/s 35 ▂▄__ WPA2

-- Infra 1 54 Mbit/s 30 ▂___ WPA1 WPA2

DIRECT-PG-FireTV_3568 Infra 153 54 Mbit/s 30 ▂___ WPA2

-- Infra 1 54 Mbit/s 29 ▂___ WPA1 WPA2 802.1X

-- Infra 1 54 Mbit/s 27 ▂___

OutOfService Infra 11 54 Mbit/s 27 ▂___ WPA2 802.1X

-- Infra 161 54 Mbit/s 27 ▂___ WPA2

Axify Infra 11 54 Mbit/s 25 ▂___ WPA2

sergek5 Infra 36 54 Mbit/s 25 ▂___ WPA1 WPA2

xfinitywifi Infra 36 54 Mbit/s 24 ▂___

XFINITY Infra 36 54 Mbit/s 24 ▂___ WPA2 802.1X

-- Infra 36 54 Mbit/s 24 ▂___ WPA1 WPA2

FiOS-A5O2K-5G Infra 161 54 Mbit/s 24 ▂___ WPA2

-- Infra 36 54 Mbit/s 22 ▂___ WPA1 WPA2 802.1X

choose the wifi you want to connect to, and run the following commend:

(base) xiaoxu@chuan:~$ nmcli d wifi connect ** password **

1

(base) xiaoxu@chuan:~$ nmcli d wifi connect ** password **

Check whether the wifi is connected:

(base) xiaoxu@chuan:~$ nmcli d
DEVICE  TYPE      STATE        CONNECTION  
wlp4s0  wifi      connected    NETGEAR91 1 
enp5s0  ethernet  unavailable  --          
lo      loopback  unmanaged    --

(base) xiaoxu@chuan:~$ nmcli d

DEVICE TYPE STATE CONNECTION

wlp4s0 wifi connected NETGEAR91 1

enp5s0 ethernet unavailable --

lo loopback unmanaged --

Notice: Actually in my desktop I still see an icon as no connection, but actually it is connected!

Numpy Precision

value = 0.0205862828046875
print("%.16f" % value)
value = value.astype(np.float32)
print("%.16f" % value)

value = 0.0205862828046875

print("%.16f" % value)

value = value.astype(np.float32)

print("%.16f" % value)

#output
0.0205862828046875
0.0205862820148468

#output

0.0205862828046875

0.0205862820148468

Precision changed!

Failed again with CUDA…

Another sad experience with cuda.

Tensorflow compiling with cuda just doesn’t work when I suspend my machine. (Error: GPU cannot be found.)
1. Tried to reinstall tensorflow again…FAILED!
2. Tried to restart the PC…WORKED!
However, I met this error again:https://github.com/zengarden/light_head_rcnn/issues/9
1. Tried to change:
  1. /home/xiaoxu/Documents/tf_install/venv/lib/python3.6/site-packages/tensorflow/include/tensorflow/core/util/cuda_device_functions.h
    1. line 32:
      1. -#”cuda/include/cuda.h”
      2. +#include “cuda.h”
  2. /home/xiaoxu/Documents/tf_install/venv/lib/python3.6/site-packages/tensorflow/include/tensorflow/core/util/cuda_kernel_helper.h
    1. line 24:
      1. -#”cuda/include/cuda_fp16.h”
      2. +#include “cuda_fp16.h”
Then, I recompiled cuda functions, and got all-zero outputs.
1. I forgot to switch cuda9.0(default) to cuda 10.0, switch to cuda 10.0…WORKED!

Switch version of g++ & Switch version of CUDA

Switch version of g++

- Example: install g++ 5.3 and g++ 7.3, then switch between them
- Step 1: install g++ 5.3 with priority 20

sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-5.3 20

1	sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-5.3 20

- Step 2: install g++ 7.3 with priority 60

sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-7.3 60

1	sudo update-alternatives --install /usr/bin/g++ g++ /usr/bin/g++-7.3 60

- Choose one g++ version:

sudo update-alternatives --config gcc
// You will see the following:
There are 2 choices for the alternative gcc (providing /usr/bin/gcc).

  Selection    Path            Priority   Status
------------------------------<wbr />------------------------------
* 0            /usr/bin/gcc-7   60        auto mode
  1            /usr/bin/gcc-5   60        manual mode
  2            /usr/bin/gcc-7   60        manual mode

Press &lt;enter&gt; to keep the current choice[*], or type selection number:

sudo update-alternatives --config gcc

// You will see the following:

There are 2 choices for the alternative gcc (providing /usr/bin/gcc).

Selection Path Priority Status

------------------------------<wbr />------------------------------

* 0 /usr/bin/gcc-7 60 auto mode

1 /usr/bin/gcc-5 60 manual mode

2 /usr/bin/gcc-7 60 manual mode

Press <enter> to keep the current choice[*], or type selection number:

Switch version of CUDA
- Download https://github.com/phohenecker/switch-cuda
- run the .sh file