"Segmentation fault (core dumped)"

[jgoenetxea]

I have followed all the steps mentioned in the readme file, but when I run the script:

th main.lua

The console says: 'Segmentation fault (core dumped)'

I have the models and the dataset included in the same folder, and trying to follow the paths described in the 'opts.lua' file.

Is there any thing more I have to do to put the program working?

[1adrianb]

@jgoenetxea I just made some changes a hour ago. Could you please check it now? Thank you.

[jgoenetxea]

I have downloaded all the changes (git pull) and I have reinstalled all elements again to be sure they are correctly installed. However, it looks like there are some 'random' errors. Y tried to run the 'main.lua' file four times with four different responses. I have never used Lua before, so I do not know what could be happening.

Here how the terminal look:

(p2k) jgoenetxea@Trantor:~/git/2D-and-3D-face-alignment$ th main.lua 
Segmentation fault (core dumped)
(p2k) jgoenetxea@Trantor:~/git/2D-and-3D-face-alignment$ th main.lua 
Fatal Python error: PyEval_SaveThread: NULL tstate
Aborted (core dumped)
(p2k) jgoenetxea@Trantor:~/git/2D-and-3D-face-alignment$ th main.lua 
Fatal Python error: ceval: tstate mix-up
Fatal Python error: GC object already tracked
Aborted (core dumped)
(p2k) jgoenetxea@Trantor:~/git/2D-and-3D-face-alignment$ th main.lua 
Fatal Python error: PyEval_SaveThread: NULL tstate
Aborted (core dumped)
(p2k) jgoenetxea@Trantor:~/git/2D-and-3D-face-alignment$ 

Any suggestion?

[jgoenetxea]

I have updated the text in the previous comment, but I don't know it the system notifies the edits.

[1adrianb]

It looks like it doesn't like the python/numpy/matplotlib versions, I remember it working with the version that is present in the ubuntu repository, but not with the ones updated via pip. Could you please run it by disabling the plotting part of the code (which uses python) or install the numpy&matplotlib versions available in the ubuntu repository ? If this doesnt help please let me know and I will have a second look (and probably release a Docker image)

[jgoenetxea]

Ok, I have commented all the 'py.exec' blocks, and it looks that the segmentation fault is fixed.

However, it asks for new dependencies not listed in the readme file: cunn, matio and npy4th

I have included them with:

luarocks install cunn
luarocks install matio
luarocks install npy4th

After those installation, is not able to find the data

Scanning directory for data...  
Found 0 **images**

, but there is not any segmentation fault.

[1adrianb]

I will add cunn to the list, I missed it somehow. The matio and npy4th should be optional and they are used if the input data is in .mat or .npy/npz format

[1adrianb]

I will also prepare a Docker image over the weekend to avoid this kind of issues in the future

[happyfrank]

I got the same issue.I commended the 'py.exec' blocks. there is not any segmentation fault. But I got this error:

Scanning directory for data...  
Found 5 images  
5 images require a face detector    
Initialising python libs... 
Initialising detector...    
/home/ubuntu/usr/local/torch/install/bin/luajit: ./facedetection_dlib.lua:22: Python error: execute Python code
Traceback (most recent call last):
  File "<string>", line 1, in <module>
NameError: name 'np' is not defined

stack traceback:
    [C]: in function 'reval'
    ./facedetection_dlib.lua:22: in function 'detect'
    main.lua:51: in main chunk
    [C]: in function 'dofile'
    ...ocal/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk
    [C]: at 0x00405d50

[1adrianb]

Hi @happyfrank, In the meantime python dlib is used for face detection and further python code removals are required. You can sanitize this and remove also this part. Alternatively you can use the Docker image provided, which should set up the environment for you. There is also another version of this code in python which should be hopefully easier to setup.

[kzw5309]

I tried to run AaronJackson/vrn that contains your 2D-and-3D-face-alignment, but I got the following error after following the solution above. Scanning directory for data... Found 5 imag Scanning directory for data...
Found 5 images
5 images require a face detector
Initialising python libs... Initialising detector...
/home/ron/usr/local/torch/install/bin/luajit: main.lua:51: Python code must be string stack traceback: [C]: in function 'detect' main.lua:51: in main chunk [C]: in function 'dofile' ...ocal/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x00405d50 es 5 images require a face detector Initialising python libs... Initialising detector... /home/ron/usr/local/torch/install/bin/luajit: main.lua:51: Python code must be string stack traceback: [C]: in function 'detect' main.lua:51: in main chunk [C]: in function 'dofile' ...ocal/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x00405d50 Do you have any ideas how to solve this problem?

[1adrianb]

Hi @kzw5309 the above solution is indeed for people that don't need the face detection, while you actually needed. You have to makr sure all the python stack is properly installed. Please check the docker image for an example.

[kzw5309]

If I uncomment python code, I will get the following error.

./run.sh: line 30: 16055 Segmentation fault (core dumped) th main.lua -model 2D-FAN-300W.t7 -input ../$INPUT/ -detectFaces true -mode generate -output ../$INPUT/ -device gpu -outputFormat txt ls: cannot access '.txt': No such file or directory ls: cannot access '.raw': No such file or directory

I have installed all dependencies you mentioned, but I don't have any experience with docker. Could you share me with more details what I should do with docker image example?

[crazySyaoran]

Hi @kzw5309 I also come from AaronJackson/vrn and got same problem with you. Did you solve it finally?

[leeeeeeo]

@happyfrank @1adrianb Hello, I met the same problem. Any suggestions please?