segmentation fault running the textdetection.py under text module

[MinghuaJiang]

System information (version)

• OpenCV => 3.1
• Operating System / Platform => Ubuntu 16.04

Detailed description

I tried to run the sample textdetection.py, but notice this line of code would cause segment fault on my computer no matter what kind of image I use.
regions = cv2.text.detectRegions(channel,er1,er2)

p.s. Something not related to this issue. I am curious why the er1 and er2 got initialized within the loop. Can I put it outside the loop?

Steps to reproduce

[alalek]

3.1

You should check the latest master version first.

would cause segment fault on my computer

Stack-trace is required.

[MinghuaJiang]

sorry, I did use the latest version on the master, which should be 3.1.0-dev.

[MinghuaJiang]

In my case, when I run this sample code, I have totally 9 channels for my image, and I tried each channel one by one, for some channel, that line of code would give me segmentation fault directly, and for other channel, it gives me this error.

*** Error in `python': double free or corruption (!prev): 0x0000000001e87450 ***
I'm not sure what does this mean.
I'm using Python 2.7.

[iesahin]

I run samples/textdetection.py with gdb and get the following stack trace. It seems a fault about findContours call in cv::text::detectRegions

Extracting Class Specific Extremal Regions from 9 channels ...
    (...) this may take a while (...)

Thread 1 "python3" received signal SIGSEGV, Segmentation fault.
__GI___libc_free (mem=0x82250a40) at malloc.c:2949
2949    malloc.c: No such file or directory.
(gdb) bt
#0  __GI___libc_free (mem=0x82250a40) at malloc.c:2949
#1  0x00007fffee854399 in cv::Mat::deallocate() () from /home/iesahin/Software/opencv/build-master/lib/libopencv_core.so.3.1
#2  0x00007fffefb7690c in cv::findContours(cv::_InputOutputArray const&, cv::_OutputArray const&, cv::_OutputArray const&, int, int, cv::Point_<int>) () from /home/iesahin/Software/opencv/build-master/lib/libopencv_imgproc.so.3.1
#3  0x00007ffff3c84d67 in cv::text::detectRegions(cv::_InputArray const&, cv::Ptr<cv::text::ERFilter> const&, cv::Ptr<cv::text::ERFilter> const&, std::vector<std::vector<cv::Point_<int>, std::allocator<cv::Point_<int> > >, std::allocator<std::vector<cv::Point_<int>, std::allocator<cv::Point_<int> > > > >&) ()
   from /home/iesahin/Software/opencv/build-master/lib/libopencv_text.so.3.1
#4  0x00007ffff678aecc in pyopencv_cv_text_detectRegions(_object*, _object*, _object*) ()
   from /home/iesahin/Environments/dervaze/lib/python3.5/site-packages/cv2.cpython-35m-x86_64-linux-gnu.so
#5  0x00000000004e1757 in PyCFunction_Call ()
#6  0x000000000051aed4 in PyEval_EvalFrameEx ()
#7  0x0000000000523de7 in ?? ()
#8  0x0000000000524b0f in PyEval_EvalCode ()
#9  0x00000000005fd5f2 in ?? ()
#10 0x00000000005ffb2a in PyRun_FileExFlags ()
#11 0x00000000005ffd1c in PyRun_SimpleFileExFlags ()
#12 0x000000000063f886 in Py_Main ()
#13 0x00000000004d00e1 in main ()

[iesahin]

Note that I'm able to run cv2.findContours standalone (without segfault) so this is likely about text module.

[imenmelkibhm]

I got a similar error when running detect-er-chars.py:

[New Thread 0x7fffc6663700 (LWP 15633)]
[New Thread 0x7fffc6262700 (LWP 15634)]
[New Thread 0x7fffc5e61700 (LWP 15635)]
[New Thread 0x7fffc5a60700 (LWP 15636)]
[New Thread 0x7fffc565f700 (LWP 15637)]
[New Thread 0x7fffc525e700 (LWP 15638)]
*** Error in `/usr/bin/python': double free or corruption (!prev): 0x00000000018d5370 ***

Program received signal SIGABRT, Aborted.
0x00007ffff6f26067 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
56 ../nptl/sysdeps/unix/sysv/linux/raise.c: No such file or directory.
(gdb) bt
#0 0x00007ffff6f26067 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:56
#1 0x00007ffff6f27448 in __GI_abort () at abort.c:89
#2 0x00007ffff6f641b4 in __libc_message (do_abort=do_abort@entry=1,
fmt=fmt@entry=0x7ffff7059210 "*** Error in `%s': %s: 0x%s \n")
at ../sysdeps/posix/libc_fatal.c:175
#3 0x00007ffff6f6998e in malloc_printerr (action=1,
str=0x7ffff7059318 "double free or corruption (!prev)", ptr=) at malloc.c:4996
#4 0x00007ffff6f6a696 in int_free (av=, p=, have_lock=0)
at malloc.c:3840
#5 0x00007ffff5befc79 in cv::Mat::deallocate() () from /usr/local/lib/libopencv_core.so.3.1
#6 0x00007ffff406e7bc in cv::findContours(cv::InputOutputArray const&, cv::OutputArray const&, cv::OutputArray const&, int, int, cv::Point) () from /usr/local/lib/libopencv_imgproc.so.3.1
#7 0x00007ffff062438e in cv::text::detectRegions(cv::InputArray const&, cv::Ptrcv::text::ERFilter const&, cv::Ptrcv::text::ERFilter const&, std::vector<std::vector<cv::Point, std::allocator<cv::Point > >, std::allocator<std::vector<cv::Point, std::allocator<cv::Point > > > >&) ()
from /usr/local/lib/libopencv_text.so.3.1
#8 0x00007ffff6a37e44 in pyopencv_cv_text_detectRegions(_object, _object, _object) ()
from /usr/local/lib/python2.7/dist-packages/cv2.so
#9 0x00000000004c9196 in PyEval_EvalFrameEx ()
#10 0x00000000004c7a59 in PyEval_EvalCodeEx ()
#11 0x000000000050160f in ?? ()
#12 0x00000000004f6c32 in PyRun_FileExFlags ()
#13 0x00000000004f5d37 in PyRun_SimpleFileExFlags ()
#14 0x00000000004981cd in Py_Main ()
#15 0x00007ffff6f12b45 in __libc_start_main (main=0x497c60

, argc=3, argv=0x7fffffffe358,
init=, fini=, rtld_fini=, stack_end=0x7fffffffe348)
at libc-start.c:287
#16 0x0000000000497b8b in _start ()
(gdb)

[imenmelkibhm]

Problem solved with this fix. However I still have this problem when running the textDetection.py (with different images):

python ./textdetection.py scenetext02.jpg

textdetection.py
A demo script of the Extremal Region Filter algorithm described in:
Neumann L., Matas J.: Real-Time Scene Text Localization and Recognition, CVPR 2012

Extracting Class Specific Extremal Regions from 9 channels ...
(...) this may take a while (...)
OpenCV Error: Assertion failed (0 <= roi.x && 0 <= roi.width && roi.x + roi.width <= m.cols && 0 <= roi.y && 0 <= roi.height && roi.y + roi.height <= m.rows) in Mat, file /opt/exe/OpenCV/modules/core/src/matrix.cpp, line 517
Traceback (most recent call last):
File "./textdetection.py", line 47, in
rects = cv2.text.erGrouping(img,channel,[r.tolist() for r in regions])
cv2.error: /opt/exe/OpenCV/modules/core/src/matrix.cpp:517: error: (-215) 0 <= roi.x && 0 <= roi.width && roi.x + roi.width <= m.cols && 0 <= roi.y && 0 <= roi.height && roi.y + roi.height <= m.rows in function Mat

[alalek]

@imenmelkibhm Thanks for report! Please grab fixes from #876 too and check again.

[imenmelkibhm]

to solve the problem of "Default classifier file not found" you can
1- At line 20 of textdetection.py put: os.path.dirname(os.path.abspath(sys.argv[0]))

2- in the command line use ./textdetection.py rather than textdetection.py when calling the python script.