about test time

[yyytq]

thank you very much for your code sharing. Is it possible to tell me the test time of the three levels? or can you give a proportion of the time cost of different level? thx again!!!

[luoyetx]

you can test the model via python2.7 test/test.py 1 for level-1 and 2,3 for level-2 & level-3, it will give you error and speed test over test dataset. And don't forget to set the model path correctly in common/cnns.py

[yyytq]

yeah. i test it. but i don't know if there is something wrong, the whole time for the network is about 1+ s per image, which i think is too long. i wonder if you ever saved the result of the test time for each level?

[luoyetx]

I test the model on a 2.2 GHz CPU, it doesn't perform that bad, Actually I got the result below.

level1: 218fps
level2: 80fps
level3; 50fps

[yyytq]

thank you very much for your sharing! ！！ i will check my work again~~(>_<)~~

[sshhuu110]

@luoyetx First of all thanks for sharing the code.

Can this code run in Linux OS Ubuntu?

When I was running this code, I got below error

Traceback (most recent call last): File "dataset/level1.py", line 16, in from common import shuffle_in_unison_scary, logger, createDir, processImage ImportError: No module named common

Please suggest how to run this code in Ubuntu? Thanks in advance !!!

[luoyetx]

@sshhuu110 I'm using CentOS but I don't think the OS is the problem. I think you run your script like python2.7 level1.py, but you should run the script under the root directory of the project like python2.7 dataset/level1.py. You can refer to this scipt bootstrap.sh.

[sshhuu110]

@luoyetx Thanks, The code is running. But how does it take to run on a CPU. What is the training time?

[luoyetx]

all prototxt files are generate by python code generate.py. The default device is GPU.

About the training time. It really depends on the GPU. And you should tuning the CNNs carefully which can cost much time.

[sshhuu110]

I changed it to CPU, I started running on Friday and it is still running. Any suggestions, is it normal training time?

[luoyetx]

I don't recommend you to use CPU which will be very slow. What's more, you should change the max_iter in solver.prototxt.template under prototxt. The default max_iter is very large and you may run many days under CPU :joy:

[yyytq]

@luoyetx @sshhuu110 sorry to interrupt you . I also have the same problem when run in Linux T^T i just use the script as bootstrap.sh. do your guys have any suggestions?

[sshhuu110]

@yyytq Now, I could able to run it on Linux OS and with only CPU.

What is the exact error are u facing?

[yyytq]

@sshhuu110 just the first error you meet.

File "dataset/level1.py", line 16, in from common import shuffle_in_unison_scary, logger, createDir, processImage ImportError: No module named common

i have changed the script, but the error remain T^T

[luoyetx]

@yyytq you may consider to modify PYTHONPATH environment. I don't know why you encounter this problem but I have no error on CentOS.

[sshhuu110]

@yyytq I did some thing like this export PYTHONPATH=$HOME/dirWithScripts/:$PYTHONPATH export PYTHONPATH=$HOME/common/:$PYTHONPATH

[sshhuu110]

@yytq Since these functions are in utils.py and utils.py is in common folder. I think we should also add common folder path to PYTHONPATH. This will solve your problem :)

[sshhuu110]

@yyytq

I think this must be dataset related problem, either you don't have data sets or train and test data set may not be set to correct paths. Check once again

On Tue, Jan 26, 2016 at 8:45 PM, yyytq notifications@github.com wrote:

@sshhuu110 https://github.com/sshhuu110 Thank you very much...but other problem occurs......😂😂

File "dataset/level1.py", line168, in generate('dataset/train/trainImageList.txt', 'train',argument=True) File"dataset/level1.py",line38,in generate Image=cv2.imread(imagepath, cv2.CV_LOAD_IMAGE_GRAYSCALE) AttributeError : 'module ' object has no attribute 'CV_LOAD_IMAGE_GRAYSCALE'

I tried to change cv2.CV_LOAD_IMAGE_GRAYSCALE to '0' or cv2.IMREAD_GRAYSCALE but didn't work😭

— Reply to this email directly or view it on GitHub https://github.com/luoyetx/deep-landmark/issues/1#issuecomment-174996283 .

[yyytq]

@sshhuu110 i'm sure the path is right.....

As i also encounter the error that can't found caffe .. i add the caffe path before 'import caffe' so that it can run did you also encounter this problem? if not…… I'll thinking if it is the problem of my software environment.....

[sshhuu110]

@yytq I have faced the "can't found caffe" problem also.

For that in level.py file in train folder you have to change cmd path to your caffe path like below

cmds = []
for t in models[level-1]:
    cmd = 'mkdir model/{0}_{1}'.format(level, t)
    os.system(cmd)
    cmd = '/home/username/Desktop/caffe-master/build/tools/caffe train --solver prototxt/{0}_{1}_solver.prototxt'.format(level, t)
    # w(os.system(cmd))
    cmds.append('/home/username/Desktop/caffe-master/build/tools/caffe train --solver prototxt/{0}_{1}_solver.prototxt'.format(level, t))

[yyytq]

@sshhuu110 yeah . i have solved this by myself.... but still troubled by the 'CV_LOAD_IMAGEGRAYSCALE' T^T can you tell me that if you do some other changes except those mentioned below? so that i can make my mental preparations beforehand ~~o(><)o ~~ or……if it is possible to send me the files that can run in Ubuntu directly?

[sshhuu110]

@yytq I had not got this "CV_LOAD_IMAGE_GRAYSCALE" problem

I did only the changes I have mentioned

attached level1 level2 level3 level.py files FYR

On Wed, Jan 27, 2016 at 3:56 PM, yyytq notifications@github.com wrote:

@sshhuu110 https://github.com/sshhuu110 yeah . i have solved this by myself.... but still troubled by the 'CV_LOAD_IMAGEGRAYSCALE' T^T can you tell me that if you do some other changes except those mentioned below? so that i can make my mental preparations beforehand ~~o(><)o ~~ or……if it is possible to send me the files that can run in Ubuntu directly?

— Reply to this email directly or view it on GitHub https://github.com/luoyetx/deep-landmark/issues/1#issuecomment-175470371 .

[yyytq]

@sshhuu110 okay!thank you a lot !!

[pidanxhq]

@sshhuu110 hi ,I have faced the "can't found caffe" problem, and i changed code as below cmd = '/home/data/caffe-master/tools/caffe train --solver prototxt/{0}_{1}_solver.prototxt'.format(level, t)

w(os.system(cmd))

    cmds.append('/home/data/caffe-master/tools/caffe train --solver prototxt/{0}_{1}_solver.prototxt'.format(level, t))

it showed "sh: 1:"/home/data/caffe-master/tools/caffe: not found",do you know how to solve this problem

[huinsysu]

@yyytq hi,have you solved the problem "'module' object has no attribute 'CV_LOAD_IMAGE_GRAYSCALE'"?I meet it too.Thanks a lot!

[huinsysu]

@yyytq Hi,I have solved it! Just replace "CV_LOAD_IMAGE_GRAYSCALE" in the code with "IMREAD_GRAYSCALE".