cannot find -lopencv_dep_cudart

[ferhatm]

HEY!
While I''m trying to compile your code with catkim_make, I'm taking this error :

Scanning dependencies of target single_image_test
[ 9%] Linking CXX executable /home/nvidia/ip_ws/devel/lib/darknet_ros/yolo_ros
[ 9%] Building CXX object darknet_ros/CMakeFiles/single_image_test.dir/src/single_image_test.cpp.o
/usr/bin/ld: cannot find -lopencv_dep_cudart
collect2: error: ld returned 1 exit status
darknet_ros/CMakeFiles/yolo_ros.dir/build.make:4557: recipe for target '/home/nvidia/ip_ws/devel/lib/darknet_ros/yolo_ros' failed
make[2]: *** [/home/nvidia/ip_ws/devel/lib/darknet_ros/yolo_ros] Error 1
CMakeFiles/Makefile2:1141: recipe for target 'darknet_ros/CMakeFiles/yolo_ros.dir/all' failed
make[1]: *** [darknet_ros/CMakeFiles/yolo_ros.dir/all] Error 2
make[1]: *** Waiting for unfinished jobs....
/home/nvidia/ip_ws/src/darknet_ros/src/single_image_test.cpp:26:13: warning: deprecated conversion from string constant to ‘char*’ [-Wwrite-strings]
char cfg = "/home/catkin_ws/src/darknet_ros/cfg/tiny-yolo-voc.cfg";
^
/home/nvidia/ip_ws/src/darknet_ros/src/single_image_test.cpp:27:17: warning: deprecated conversion from string constant to ‘char’ [-Wwrite-strings]
char *weights = "/home/catkin_ws/src/darknet_ros/weights/tiny-yolo-voc.weights";
^
[ 9%] Linking CXX executable /home/nvidia/ip_ws/devel/lib/darknet_ros/single_image_test
/usr/bin/ld: cannot find -lopencv_dep_cudart
collect2: error: ld returned 1 exit status
darknet_ros/CMakeFiles/single_image_test.dir/build.make:2900: recipe for target '/home/nvidia/ip_ws/devel/lib/darknet_ros/single_image_test' failed
make[2]: *** [/home/nvidia/ip_ws/devel/lib/darknet_ros/single_image_test] Error 1
CMakeFiles/Makefile2:1609: recipe for target 'darknet_ros/CMakeFiles/single_image_test.dir/all' failed
make[1]: *** [darknet_ros/CMakeFiles/single_image_test.dir/all] Error 2
Makefile:138: recipe for target 'all' failed
make: *** [all] Error 2
Invoking "make -j4 -l4" failed

Any idea?
Thanks

[pgigioli]

Do you have CUDA installed? Unfortunately, I haven't built CPU support yet.
What OpenCV version do you have? Make sure its updated to version 3.

If you do have CUDA installed and OpenCV3, try adding set(CUDA_USE_STATIC_CUDA_RUNTIME OFF) to your CMakeLists.txt. This might be related to an issue with your OpenCV version similar to this.

[ferhatm]

Yes, I'm trying with CUDA and my OpenCV version is 3.3.1. Tried everything on the link, the best result is :

[ 65%] Building NVCC (Device) object darknet_ros/CMakeFiles/yolo_ros.dir/darknet/src/yolo_ros_generated_maxpool_layer_kernels.cu.o
nvcc warning : The 'compute_20', 'sm_20', and 'sm_21' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
nvcc warning : The 'compute_20', 'sm_20', and 'sm_21' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
[ 67%] Linking CXX executable /home/nvidia/ip_ws/devel/lib/darknet_ros/yolo_ros
CMakeFiles/yolo_ros.dir/src/yolo_ros.cpp.o: In function yoloObjectDetector::~yoloObjectDetector()': /home/nvidia/ip_ws/src/darknet_ros/src/yolo_ros.cpp:89: undefined reference to cv::destroyWindow(cv::String const&)'
CMakeFiles/yolo_ros.dir/src/yolo_ros.cpp.o: In function yoloObjectDetector::runYOLO(cv::Mat&)': /home/nvidia/ip_ws/src/darknet_ros/src/yolo_ros.cpp:178: undefined reference to cv::imshow(cv::String const&, cv::_InputArray const&)'
CMakeFiles/yolo_ros.dir/src/yolo_ros.cpp.o: In function yoloObjectDetector::yoloObjectDetector()': /home/nvidia/ip_ws/src/darknet_ros/src/yolo_ros.cpp:84: undefined reference to cv::namedWindow(cv::String const&, int)'
collect2: error: ld returned 1 exit status
darknet_ros/CMakeFiles/yolo_ros.dir/build.make:4557: recipe for target '/home/nvidia/ip_ws/devel/lib/darknet_ros/yolo_ros' failed
make[2]: *** [/home/nvidia/ip_ws/devel/lib/darknet_ros/yolo_ros] Error 1
CMakeFiles/Makefile2:1141: recipe for target 'darknet_ros/CMakeFiles/yolo_ros.dir/all' failed
make[1]: *** [darknet_ros/CMakeFiles/yolo_ros.dir/all] Error 2
Makefile:138: recipe for target 'all' failed
make: *** [all] Error 2

while trying this cmake file :

set(CUDA_USE_STATIC_CUDA_RUNTIME OFF)
set(
CUDA_NVCC_FLAGS
${CUDA_NVCC_FLAGS};
-O3
-gencode arch=compute_20,code=[sm_20,sm_21]
-gencode arch=compute_30,code=sm_30
-gencode arch=compute_35,code=sm_35
-gencode arch=compute_50,code=[sm_50,compute_50]
-gencode arch=compute_52,code=[sm_52,compute_52]
-gencode arch=compute_53,code=sm_53
-gencode arch=compute_61,code=sm_61
-gencode arch=compute_62,code=sm_62

)

--
after typing :
sudo ln -s /usr/local/cuda/lib64/libcudart.so /usr/lib/libopencv_dep_cudart.so

Any idea again?
Thanks

[pgigioli]

Might be an issue with opencv 3.3.1 similar to this. You may have to update kinetic or fiddle around with your opencv install.

You could try building the docker image as a sanity check. The docker image uses opencv 2.4.

Let me know if you figure it out.