Ask Your Question
0

vision_ssd_detect cannot be launched (fine compilation of ssdcaffe and CUDA enabled device)

asked 2019-06-29 17:50:42 -0500

cassini.huygens gravatar image

Hi all,

It's not possible to launch vision_ssd_detect because it is not possible to find the "libcaffe.so.1.0.0-rc3". Compilation with ssdcaffe went well (followed http://caffe.berkeleyvision.org/insta...) and then make & make distribute. I am not able to track where this issue is coming from. I attach the terminal output for ./colcon_release with CUDA flag 1 for debugging purposes.

System Specs:

Ubuntu 16.04 ROS Kinetic Autoware git commit: 42aba146 CUDA 8.0

Autoware build result - SSDCAFFE build result and .launch log file: 1- https://drive.google.com/file/d/1Epki... 2- https://drive.google.com/file/d/1UPDz... 3- https://drive.google.com/file/d/1ltTW...

/home/ocp/Autoware/ros/install/runtime_manager/lib/runtime_manager/runtime_manager_dialog.py:2865: wxPyDeprecationWarning: Call to deprecated item. wx.InitAllImageHandlers() Unable to register with master node [http://localhost:11311]: master may not be running yet. Will keep trying. loading qs.yaml Subscribe[localization] topic=/ndt_stat, key=/ndt_stat.NDTStat.exe_time Subscribe[detection] topic=/topic2, key=/topic2 Subscribe[detection] topic=/topic1, key=/topic1 loading setup.yaml ['rosparam', 'set', 'tf_x', '1.2'] ['rosparam', 'set', 'tf_y', '0.0'] ['rosparam', 'set', 'tf_z', '2.0'] ['rosparam', 'set', 'tf_yaw', '0.0'] ['rosparam', 'set', 'tf_pitch', '0.0'] ['rosparam', 'set', 'tf_roll', '0.0'] ['rosparam', 'set', 'localizer', 'velodyne'] loading map.yaml loading sensing.yaml loading computing.yaml loading interface.yaml loading data.yaml loading simulation.yaml ['rosparam', 'set', '/use_sim_time', 'false'] loading status.yaml loading state.yaml loading topics.yaml ['roslaunch', 'vision_ssd_detect', 'vision_ssd_detect.launch', 'score_threshold:=0.6', 'image_src:=/image_raw', 'network_definition_file:=/home/ocp/ssdcaffe/models/VGGNet/VOC0712Plus/SSD_300x300/deploy.prototxt', 'pretrained_model_file:=/home/ocp/ssdcaffe/models/VGGNet/VOC0712Plus/SSD_300x300/VGG_VOC0712Plus_SSD_300x300_iter_240000.caffemodel', 'use_gpu:=True', 'gpu_device_id:=0'] pid=2337 sched policy=OTHER prio=0 /home/ocp/Autoware/ros/install/vision_ssd_detect/lib/vision_ssd_detect/vision_ssd_detect: error while loading shared libraries: libcaffe.so.1.0.0-rc3: cannot open shared object file: No such file or directory [vision_ssd_detect-1] process has died [pid 2356, exit code 127, cmd /home/ocp/Autoware/ros/install/vision_ssd_detect/lib/vision_ssd_detect/vision_ssd_detect __name:=vision_ssd_detect __log:=/home/ocp/.ros/log/d9212ea6-9a00-11e9-8b7c-a4fc771a0d0d/vision_ssd_detect-1.log]. log file: /home/ocp/.ros/log/d9212ea6-9a00-11e9-8b7c-a4fc771a0d0d/vision_ssd_detect-1*.log /image_raw

Thanks in advance!

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
2

answered 2019-06-29 22:33:07 -0500

amc-nu gravatar image

The error itself is telling you that the linux linker cannot found libcaffe.so.

/lib/vision_ssd_detect/vision_ssd_detect: error while loading shared libraries: libcaffe.so.1.0.0-rc3

You'll need to either 1. add the path where that .so file is located to the LD_LIBRARY_PATH, or 2. "install" or move caffe headers/libraries to the system path, and update the libraries cache.

Supponsing you cloned Caffe to /home/YOUR_USER/caffe/

If you go for 1. $ export LD_LIBRARY_PATH=/home/YOUR_USER/caffe/distribute/lib:$LD_LIBRARY_PATH

If you prefer 2. $ cd /home/YOUR_USER/caffe && sudo make install && sudo ldconfig

Then you should be able to run the node.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower

Stats

Asked: 2019-06-29 17:50:42 -0500

Seen: 46 times

Last updated: Jun 29