Seg fault running run_video_slam example

[Sid220]

I installed stella_vslam on Ubuntu 22.04 by following the docs. While running the run_video_slam example using the following command,

./run_video_slam -v ./orb_vocab.fbow -m ./aist_living_lab_1/video.mp4 -c ~/Documents/stella_vslam/example/aist/equirectangular.yaml --frame-skip 3 --no-sleep --map-db-out map.msg

I get a segmentation fault. Here is the full output:

Full output

[2024-01-07 17:02:40.548] [I] config file loaded: /home/sid/Documents/stella_vslam/example/aist/equirectangular.yaml
--start-timestamp is not set. using system timestamp.
If --no-sleep is set without --start-timestamp, timestamps may overlap between multiple runs.
[2024-01-07 17:02:40.548] [I] 
original version of OpenVSLAM,
Copyright (C) 2019,
National Institute of Advanced Industrial Science and Technology (AIST)
All rights reserved.
stella_vslam (the changes after forking from OpenVSLAM),
Copyright (C) 2022, stella-cv, All rights reserved.

This is free software,
and you are welcome to redistribute it under certain conditions.
See the LICENSE file.

Camera:
  name: RICOH THETA S 960
  setup: monocular
  model: equirectangular
  fps: 30.0
  cols: 1920
  rows: 960
  color_order: RGB
Preprocessing:
  min_size: 800
  mask_rectangles:
    - [0.0, 1.0, 0.0, 0.1]
    - [0.0, 1.0, 0.84, 1.0]
    - [0.0, 0.2, 0.7, 1.0]
    - [0.8, 1.0, 0.7, 1.0]
Feature:
  name: default ORB feature extraction setting
  scale_factor: 1.2
  num_levels: 8
  ini_fast_threshold: 20
  min_fast_threshold: 7
Mapping:
  backend: g2o
  baseline_dist_thr_ratio: 0.02
  redundant_obs_ratio_thr: 0.95
  num_covisibilities_for_landmark_generation: 20
  num_covisibilities_for_landmark_fusion: 20
Tracking:
  backend: g2o
LoopDetector:
  backend: g2o
  enabled: true
  reject_by_graph_distance: true
  min_distance_on_graph: 50
System:
  map_format: msgpack

[2024-01-07 17:02:40.548] [I] loading ORB vocabulary: ./orb_vocab.fbow
[2024-01-07 17:02:40.585] [I] load orb_params "default ORB feature extraction setting"
[2024-01-07 17:02:40.587] [I] startup SLAM system
[2024-01-07 17:02:40.587] [I] start mapping module
[2024-01-07 17:02:40.588] [I] start global optimization module
Stack trace (most recent call last) in thread 82060:
#17   Object "", at 0xffffffffffffffff, in 
#16   Source "./misc/../sysdeps/unix/sysv/linux/x86_64/clone3.S", line 81, in __clone3 [0x7fdbd212665f]
#15   Source "./nptl/pthread_create.c", line 442, in start_thread [0x7fdbd2094ac2]
#14   Object "/lib/x86_64-linux-gnu/libstdc++.so.6", at 0x7fdbd24dc252, in std::error_code::default_error_condition() const
Gtk-Message: 17:02:40.766: Failed to load module "canberra-gtk-module"
#13 | Source "/usr/include/c++/11/bits/std_thread.h", line 211, in _M_run
    |   210: 	void
    | > 211: 	_M_run() { _M_func(); }
    |   212:       };
    | Source "/usr/include/c++/11/bits/std_thread.h", line 266, in operator()
    |   264: 	  using _Indices
    |   265: 	    = typename _Build_index_tuple<tuple_size<_Tuple>::value>::__type;
    | > 266: 	  return _M_invoke(_Indices());
    |   267: 	}
    |   268:       };
    | Source "/usr/include/c++/11/bits/std_thread.h", line 259, in _M_invoke<0>
    |   257: 	  typename __result<_Tuple>::type
    |   258: 	  _M_invoke(_Index_tuple<_Ind...>)
    | > 259: 	  { return std::__invoke(std::get<_Ind>(std::move(_M_t))...); }
    |   260: 
    |   261: 	typename __result<_Tuple>::type
    | Source "/usr/include/c++/11/bits/invoke.h", line 96, in __invoke<mono_tracking(const std::shared_ptr<stella_vslam::system>&, const std::shared_ptr<stella_vslam::config>&, const string&, const string&, unsigned int, unsigned int, bool, bool, bool, const string&, const string&, double, const string&)::<lambda()> >
    |    94:       using __type = typename __result::type;
    |    95:       using __tag = typename __result::__invoke_type;
    | >  96:       return std::__invoke_impl<__type>(__tag{}, std::forward<_Callable>(__fn),
    |    97: 					std::forward<_Args>(__args)...);
    |    98:     }
    | Source "/usr/include/c++/11/bits/invoke.h", line 61, in __invoke_impl<void, mono_tracking(const std::shared_ptr<stella_vslam::system>&, const std::shared_ptr<stella_vslam::config>&, const string&, const string&, unsigned int, unsigned int, bool, bool, bool, const string&, const string&, double, const string&)::<lambda()> >
    |    59:     constexpr _Res
    |    60:     __invoke_impl(__invoke_other, _Fn&& __f, _Args&&... __args)
    | >  61:     { return std::forward<_Fn>(__f)(std::forward<_Args>(__args)...); }
    |    62: 
    |    63:   template<typename _Res, typename _MemFun, typename _Tp, typename... _Args>
      Source "/home/sid/Documents/stella_vslam/stella_vslam_examples/src/run_video_slam.cc", line 105, in std::thread::_State_impl<std::thread::_Invoker<std::tuple<mono_tracking(std::shared_ptr<stella_vslam::system> const&, std::shared_ptr<stella_vslam::config> const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, unsigned int, unsigned int, bool, bool, bool, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, double, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&)::{lambda()#1}> > >::_M_run() [0x5650873b5f07]
        103:             if (!frame.empty() && (num_frame % frame_skip == 0)) {
        104:                 // input the current frame and estimate the camera pose
      > 105:                 slam->feed_monocular_frame(frame, timestamp, mask);
        106:             }
        107: 
        108:             const auto tp_2 = std::chrono::steady_clock::now();
#12   Source "/home/sid/Documents/stella_vslam/src/stella_vslam/system.cc", line 447, in stella_vslam::system::feed_monocular_frame(cv::Mat const&, double, cv::Mat const&) [0x7fdbd3ec7760]
        444:         spdlog::warn("preprocess: empty image");
        445:         return nullptr;
        446:     }
      > 447:     return feed_frame(create_monocular_frame(img, timestamp, mask), img);
        448: }
        449: 
        450: std::shared_ptr<Mat44_t> system::feed_stereo_frame(const cv::Mat& left_img, const cv::Mat& right_img, const double timestamp, const cv::Mat& mask) {
#11   Source "/home/sid/Documents/stella_vslam/src/stella_vslam/system.cc", line 293, in stella_vslam::system::create_monocular_frame(cv::Mat const&, double, cv::Mat const&) [0x7fdbd3ec58ba]
        291:     // Extract ORB feature
        292:     keypts_.clear();
      > 293:     extractor_left_->extract(img_gray, mask, keypts_, frm_obs.descriptors_);
        294:     frm_obs.num_keypts_ = keypts_.size();
        295:     if (keypts_.empty()) {
        296:         spdlog::warn("preprocess: cannot extract any keypoints");
#10   Source "/home/sid/Documents/stella_vslam/src/stella_vslam/feature/orb_extractor.cc", line 35, in stella_vslam::feature::orb_extractor::extract(cv::_InputArray const&, cv::_InputArray const&, std::vector<cv::KeyPoint, std::allocator<cv::KeyPoint> >&, cv::_OutputArray const&) [0x7fdbd3f700dc]
         32:     assert(image.type() == CV_8UC1);
         33: 
         34:     // build image pyramid
      >  35:     compute_image_pyramid(image);
         36: 
         37:     // mask initialization
         38:     if (!mask_is_initialized_ && !mask_rects_.empty()) {
#9    Source "/home/sid/Documents/stella_vslam/src/stella_vslam/feature/orb_extractor.cc", line 124, in stella_vslam::feature::orb_extractor::compute_image_pyramid(cv::Mat const&) [0x7fdbd3f6dbf6]
        121:         const double scale = orb_params_->scale_factors_.at(level);
        122:         const cv::Size size(std::round(image.cols * 1.0 / scale), std::round(image.rows * 1.0 / scale));
        123:         // resize
      > 124:         cv::resize(image_pyramid_.at(level - 1), image_pyramid_.at(level), size, 0, 0, cv::INTER_LINEAR);
        125:     }
        126: }
#8    Object "/usr/local/lib/libopencv_imgproc.so.3.4", at 0x7fdbcfd05739, in cv::resize(cv::_InputArray const&, cv::_OutputArray const&, cv::Size_<int>, double, double, int)
#7    Source "resize.cpp", line 0, in cv::errorNoReturn(int, cv::String const&, char const*, char const*, int) [0x7fdbcfcd72da]
#6    Object "/usr/local/lib/libopencv_core.so.3.4", at 0x7fdbcecec53a, in cv::error(int, cv::String const&, char const*, char const*, int)
#5    Object "/usr/local/lib/libopencv_core.so.3.4", at 0x7fdbcecec465, in cv::Exception::Exception(int, cv::String const&, cv::String const&, cv::String const&, int)
#4    Object "/usr/local/lib/libopencv_core.so.405", at 0x7fdbd2cf10ec, in cv::Exception::formatMessage()
#3    Object "/usr/local/lib/libopencv_core.so.405", at 0x7fdbd2cf182e, in cv::format[abi:cxx11](char const*, ...)
#2    Source "./libio/vsnprintf.c", line 114, in __vsnprintf_internal [0x7fdbd2088499]
#1    Source "./stdio-common/vfprintf-internal.c", line 1517, in __vfprintf_internal [0x7fdbd2076d30]
#0    Source "./string/../sysdeps/x86_64/multiarch/strlen-evex.S", line 77, in __strlen_evex [0x7fdbd21b1ebc]
Segmentation fault (Address not mapped to object [0xf59])
[1]    82049 segmentation fault (core dumped)  ./run_video_slam -v ./orb_vocab.fbow -m ./aist_living_lab_1/video.mp4 -c   3 

The windows open and the one with buttons seems mostly responsive, with the exception of the Terminate button. Then, after about 20 seconds, they close.

I'm not sure if I'm doing something wrong or if there is a bug. In order to build stella_vslam I had to turn off aruco detection in CMakeLists.txt:

+ set(USE_ARUCO OFF CACHE BOOL "Enable aruco")
- set(USE_ARUCO ON CACHE BOOL "Enable aruco")

Maybe this has something to do with it?

[ymd-stella]

According to the stack trace, this is due to multiple versions of OpenCV being installed.

#5    Object "/usr/local/lib/libopencv_core.so.3.4", at 0x7fdbcecec465, in cv::Exception::Exception(int, cv::String const&, cv::String const&, cv::String const&, int)
#4    Object "/usr/local/lib/libopencv_core.so.405", at 0x7fdbd2cf10ec, in cv::Exception::formatMessage()

[Sid220]

That's the issue, thanks!