Segmentation Fault when invoking GetDynSeg

[y9luiz]

Hello, I'm trying to run this project and for the mask-rcnn approach I'm facing some troubles ...

A unexpected segmentation fault occurs when I try to run with the Mask-RCNN enabled and after a few debugging I found this:

the problem occurs when I call GetDynSeg at line 62 of the file src/MaskNet.cc

PyObject* py_mask_image = PyObject_CallMethod(this->net, const_cast<char*>(this->get_dyn_seg.c_str()),"(O)",py_image);

this line jumps to the python script src/python/MaskRCNN.py, more specificly to function

def GetDynSeg(self, image,image2=None)

the function expect that the second argument (image) must be a numpy array like any other python OpenCV image, but

when the code try to get the content from the image it will produce a segmentation fault

My conclusion was that numpy array is not been correctly serialized at NDArrayConverter::toNDArray() function

[y9luiz]

the test script src/python/Check.py is working nicely, so I believe that problem is from CPP side

[y9luiz]

OS info: Ubuntu 20.04 x86_64
OpenCV Version: 3.4.7
Numpy Version: 1.21.5
g++ Version: 7.5.0

[y9luiz]

I solve this problem downgrading to opencv 2.4

[alexs7]

@y9luiz hello, have you got DynaSLAM working now ? can I ask what we should pass for the "PATH_TO_MASKS" ?

[y9luiz]

@y9luiz hello, have you got DynaSLAM working now ? can I ask what we should pass for the "PATH_TO_MASKS" ?

Yeah @alexs7 , now it's working, the parameter PATH_TO_MASKS is the folder that u want to store the masks generated by the algorithm

[alexs7]

I run this:

./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUM3.yaml ../datasets/TUM/rgbd_dataset_freiburg3_walking_rpy/ Examples/RGB-D/associations/fr3_walking_rpy.txt Examples/RGB-D/masks_temp Examples/RGB-D/masks_temp_output

and I get this:

Creating net instance...
Loading net parameters...
[ WARN:[email protected]] imread_('no_save/no_file'): can't open/read file: check file path/integrity
[1] 17264 bus error (core dumped) ./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUM3.yaml

Did you have this error ?

[y9luiz]

I run this:

./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUM3.yaml ../datasets/TUM/rgbd_dataset_freiburg3_walking_rpy/ Examples/RGB-D/associations/fr3_walking_rpy.txt Examples/RGB-D/masks_temp Examples/RGB-D/masks_temp_output

and I get this:

Creating net instance...
Loading net parameters...
[ WARN:[email protected]] imread_('no_save/no_file'): can't open/read file: check file path/integrity
[1] 17264 bus error (core dumped) ./Examples/RGB-D/rgbd_tum Vocabulary/ORBvoc.txt Examples/RGB-D/TUM3.yaml

Did you have this error ?

I have no ideia about your error.

Are you alredy tried to execute the script
src/python/Check.py ?

this script should test your python enviroment and check if is well configured

[alexs7]

@y9luiz sorry for the late reply. here is what I get when I run that script:

python2.7 src/python/Check.py
/home/alex/.local/lib/python2.7/site-packages/pycocotools/coco.py:49: UserWarning:
This call to matplotlib.use() has no effect because the backend has already
been chosen; matplotlib.use() must be called before pylab, matplotlib.pyplot,
or matplotlib.backends is imported for the first time.

The backend was originally set to 'TkAgg' by the following code:
File "src/python/Check.py", line 8, in
import matplotlib.pyplot as plt
File "/home/alex/.local/lib/python2.7/site-packages/matplotlib/pyplot.py", line 71, in
from matplotlib.backends import pylab_setup
File "/home/alex/.local/lib/python2.7/site-packages/matplotlib/backends/init.py", line 17, in
line for line in traceback.format_stack()

import matplotlib; matplotlib.use('Agg')
Using TensorFlow backend.
Initializing Mask RCNN netwFork...
./src/python
WARNING:tensorflow:From /home/alex/.local/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.
Instructions for updating:
Colocations handled automatically by placer.
WARNING:tensorflow:From /home/alex/.local/lib/python2.7/site-packages/keras/backend/tensorflow_backend.py:3144: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.
Instructions for updating:
Please use rate instead of keep_prob. Rate should be set to rate = 1 - keep_prob.
WARNING:tensorflow:From /home/alex/DynaSLAM/src/python/model.py:746: py_func (from tensorflow.python.ops.script_ops) is deprecated and will be removed in a future version.
Instructions for updating:
tf.py_func is deprecated in TF V2. Instead, use
tf.py_function, which takes a python function which manipulates tf eager
tensors instead of numpy arrays. It's easy to convert a tf eager tensor to
an ndarray (just call tensor.numpy()) but having access to eager tensors
means tf.py_functions can use accelerators such as GPUs as well as
being differentiable using a gradient tape.

2022-01-18 13:25:28.892448: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA
2022-01-18 13:25:28.912843: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 3199980000 Hz
2022-01-18 13:25:28.913257: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x56177eebd2e0 executing computations on platform Host. Devices:
2022-01-18 13:25:28.913277: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): ,
Initialated Mask RCNN network...
Mask R-CNN is correctly working

[alexs7]

@y9luiz except some warnings this shouls be OK right ? What do you get ?

[y9luiz]

@alexs7 the python environment its okay, did you are using the opencv 2.4? how I said in this topic, my issues stopped after downgrade the opencv version

[alexs7]

@y9luiz yes fixed it by downgrading to opencv 2.4

did you get the inpainting code working?

[y9luiz]

@alexs7 no I never try to use the impainting functionality

[alexs7]

@y9luiz I got my own fork here with updated details on how to build: https://github.com/alexs7/DynaSLAM

[y9luiz]

Nice job @alexs7!

[upperfish]

Hello, I'm trying to run this project and for the mask-rcnn approach I'm facing some troubles ...

A unexpected segmentation fault occurs when I try to run with the Mask-RCNN enabled and after a few debugging I found this:

the problem occurs when I call GetDynSeg at line 62 of the file src/MaskNet.cc

PyObject* py_mask_image = PyObject_CallMethod(this->net, const_cast<char*>(this->get_dyn_seg.c_str()),"(O)",py_image);

this line jumps to the python script src/python/MaskRCNN.py, more specificly to function

def GetDynSeg(self, image,image2=None)

the function expect that the second argument (image) must be a numpy array like any other python OpenCV image, but

when the code try to get the content from the image it will produce a segmentation fault

My conclusion was that numpy array is not been correctly serialized at NDArrayConverter::toNDArray() function

do you slove? i have same problem? please help god!