Missing CUDA support for opencv-python build from jetson-ai-lab (original) (raw)
April 24, 2025, 10:36am 1
I’m using a Jetson Orin Nano and have created a docker image using nvcr.io/nvidia/l4t-cuda:12.6.11-runtime as a base image and installed opencv-python from here: https://pypi.jetson-ai-lab.dev/jp6/cu126/+f/e6d/a8e91fd7e5f79/opencv_python-4.11.0-py3-none-any.whl#sha256=e6da8e91fd7e5f79509500808b38b616af4d3a8da4a8e0a3de6fa1d4da2f4fc8
However, testing gives:
$ python3
Python 3.10.12 (main, Feb 4 2025, 14:57:36) [GCC 11.4.0] on linux
Type “help”, “copyright”, “credits” or “license” for more information.
import cv2
print(cv2.cuda.getCudaEnabledDeviceCount())
0
I checked the availability of the GPU inside the container using the deviceQuery executable.
files.zip (9.3 KB)
AastaLLL April 25, 2025, 4:21am 3
Hi,
Could you double-check if you are using the installed OpenCV instead of the default one first?
>>> print(cv2.getBuildInformation())
Thanks.
naisy April 25, 2025, 7:31am 4
The container nvcr.io/nvidia/l4t-cuda:12.6.11-runtime does not include libcudnn.so.9, which is required by /usr/local/lib/python3.10/dist-packages/cv2/cv2.abi3.so.
As an alternative, you can use the nvcr.io/nvidia/deepstream-l4t:7.1-triton-multiarch container. Below is a description of how to use it.
docker run
sudo docker run --gpus all -it -v /tmp/:/tmp/ nvcr.io/nvidia/deepstream-l4t:7.1-triton-multiarch /bin/bash
install opencv
pip install \
opencv-contrib-python==4.11.0.86 \
opencv-python==4.11.0 \
--index-url https://pypi.jetson-ai-lab.dev/jp6/cu126/+simple/
check missing libs
ldd /usr/local/lib/python3.10/dist-packages/cv2/cv2.abi3.so | grep "not found"
install libs
apt update
apt install -y \
libtesseract-dev \
libtbb2 \
libmpg123-0 \
libatlas3-base \
libgtk-3-0 \
libavcodec-extra58 \
libavformat-dev \
libswscale-dev
apt install --reinstall -y \
libavutil56 \
libvpx7 \
libmp3lame0 \
libx264-163 \
libx265-199 \
libxvidcore4
check missing libs
ldd /usr/local/lib/python3.10/dist-packages/cv2/cv2.abi3.so | grep "not found"
check cv2
python3 -c "import cv2; print(cv2.cuda.getCudaEnabledDeviceCount()); print(cv2.getBuildInformation())"
1
General configuration for OpenCV 4.11.0 =====================================
Version control: 4.11.0
Extra modules:
Location (extra): /opt/opencv-python/opencv_contrib/modules
Version control (extra): 4.11.0
Platform:
Timestamp: 2025-03-05T16:32:02Z
Host: Linux 5.15.148-tegra aarch64
CMake: 3.31.6
CMake generator: Ninja
CMake build tool: /usr/local/bin/ninja
Configuration: RELEASE
Algorithm Hint: ALGO_HINT_ACCURATE
CPU/HW features:
Baseline: NEON FP16
requested: DETECT
required: NEON
Dispatched code generation: NEON_DOTPROD NEON_FP16 NEON_BF16
requested: NEON_FP16 NEON_BF16 NEON_DOTPROD
NEON_DOTPROD (1 files): + NEON_DOTPROD
NEON_FP16 (2 files): + NEON_FP16
NEON_BF16 (0 files): + NEON_BF16
C/C++:
Built as dynamic libs?: NO
C++ standard: 11
C++ Compiler: /usr/bin/c++ (ver 11.4.0)
C++ flags (Release): -fsigned-char -W -Wall -Wreturn-type -Wnon-virtual-dtor -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -fvisibility-inlines-hidden -O3 -DNDEBUG -DNDEBUG
C++ flags (Debug): -fsigned-char -W -Wall -Wreturn-type -Wnon-virtual-dtor -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -fvisibility-inlines-hidden -g -O0 -DDEBUG -D_DEBUG
C Compiler: /usr/bin/cc
C flags (Release): -fsigned-char -W -Wall -Wreturn-type -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -O3 -DNDEBUG -DNDEBUG
C flags (Debug): -fsigned-char -W -Wall -Wreturn-type -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -g -O0 -DDEBUG -D_DEBUG
Linker flags (Release): -Wl,--gc-sections -Wl,--as-needed -Wl,--no-undefined
Linker flags (Debug): -Wl,--gc-sections -Wl,--as-needed -Wl,--no-undefined
ccache: NO
Precompiled headers: NO
Extra dependencies: /usr/lib/aarch64-linux-gnu/liblapack.so /usr/lib/aarch64-linux-gnu/libcblas.so /usr/lib/aarch64-linux-gnu/libatlas.so /usr/lib/aarch64-linux-gnu/libjpeg.so /usr/lib/aarch64-linux-gnu/libpng.so /usr/lib/aarch64-linux-gnu/libz.so Iconv::Iconv m pthread cudart_static dl rt nppc nppial nppicc nppidei nppif nppig nppim nppist nppisu nppitc npps cublas cudnn cufft -L/usr/local/cuda/lib64 -L/usr/lib/aarch64-linux-gnu
3rdparty dependencies: libprotobuf ade ittnotify libwebp libtiff libopenjp2 IlmImf tegra_hal
OpenCV modules:
To be built: alphamat aruco bgsegm bioinspired calib3d ccalib core cudaarithm cudabgsegm cudacodec cudafeatures2d cudafilters cudaimgproc cudalegacy cudaobjdetect cudaoptflow cudastereo cudawarping cudev datasets dnn dnn_objdetect dnn_superres dpm face features2d flann fuzzy gapi hfs highgui img_hash imgcodecs imgproc intensity_transform line_descriptor mcc ml objdetect optflow phase_unwrapping photo plot python3 quality rapid reg rgbd saliency shape signal stereo stitching structured_light superres surface_matching text tracking video videoio videostab wechat_qrcode xfeatures2d ximgproc xobjdetect xphoto
Disabled: freetype world
Disabled by dependency: -
Unavailable: cannops cvv fastcv hdf java julia matlab ovis python2 sfm ts viz
Applications: -
Documentation: NO
Non-free algorithms: YES
GUI: GTK3
GTK+: YES (ver 3.24.33)
VTK support: NO
Media I/O:
ZLib: /usr/lib/aarch64-linux-gnu/libz.so (ver 1.2.11)
JPEG: /usr/lib/aarch64-linux-gnu/libjpeg.so (ver 80)
WEBP: build (ver decoder: 0x0209, encoder: 0x020f, demux: 0x0107)
AVIF: NO
PNG: /usr/lib/aarch64-linux-gnu/libpng.so (ver 1.6.37)
TIFF: build (ver 42 - 4.6.0)
JPEG 2000: build (ver 2.5.0)
OpenEXR: build (ver 2.3.0)
GIF: NO
HDR: YES
SUNRASTER: YES
PXM: YES
PFM: YES
Video I/O:
FFMPEG: YES
avcodec: YES (58.134.100)
avformat: YES (58.76.100)
avutil: YES (56.70.100)
swscale: YES (5.9.100)
avresample: NO
GStreamer: YES (1.20.3)
v4l/v4l2: YES (linux/videodev2.h)
Parallel framework: TBB (ver 2020.3 interface 11103)
Trace: YES (with Intel ITT)
Other third-party libraries:
Lapack: YES (/usr/lib/aarch64-linux-gnu/liblapack.so /usr/lib/aarch64-linux-gnu/libcblas.so /usr/lib/aarch64-linux-gnu/libatlas.so)
Eigen: YES (ver 3.4.0)
Custom HAL: YES (carotene (ver 0.0.1))
Protobuf: build (3.19.1)
Flatbuffers: builtin/3rdparty (23.5.9)
NVIDIA CUDA: YES (ver 12.6, CUFFT CUBLAS FAST_MATH)
NVIDIA GPU arch: 87
NVIDIA PTX archs:
cuDNN: YES (ver 9.4.0)
Python 3:
Interpreter: /usr/bin/python3.10 (ver 3.10.12)
Libraries: /usr/lib/aarch64-linux-gnu/libpython3.10.so (ver 3.10.12)
Limited API: YES (ver 0x03060000)
numpy: /tmp/pip-build-env-ytul3b_k/overlay/local/lib/python3.10/dist-packages/numpy/_core/include (ver 2.2.3)
install path: python/cv2/python-3
Python (for build): /usr/bin/python3.10
Java:
ant: NO
Java: NO
JNI: NO
Java wrappers: NO
Java tests: NO
Install to: /opt/opencv-python/_skbuild/linux-aarch64-3.10/cmake-install
-----------------------------------------------------------------
example code
vi test.py
import cv2
import os
input_file = "/opt/nvidia/deepstream/deepstream-7.1/samples/streams/sample_720p.jpg"
output_file = "/tmp/flipped_output.jpg"
# Load the image from disk (on CPU)
image = cv2.imread(input_file, cv2.IMREAD_COLOR)
if image is None:
raise RuntimeError(f"Failed to load the image file: {input_file}")
# Upload the image to GPU memory
gpu_image = cv2.cuda_GpuMat()
gpu_image.upload(image)
# Flip the image horizontally on the GPU (flipCode = 1)
gpu_flipped = cv2.cuda.flip(gpu_image, 1)
# Download the flipped image from GPU to CPU
flipped_image = gpu_flipped.download()
# Save the result as a new file
cv2.imwrite(output_file, flipped_image)
print(f"Flipped image saved to: {output_file}")
run test.py
python3 test.py
The output will be saved as /tmp/flipped_output.jpg.
Since the Docker container is started with the -v /tmp/:/tmp/ option, you can access the result directly from the Jetson Orin Nano.
This gives me the following:
\n
\n
General configuration for OpenCV 4.11.0 =====================================
Version control: 4.11.0-dirty
Extra modules:
Location (extra): /io/opencv_contrib/modules
Version control (extra): 4.11.0
Platform:
Timestamp: 2025-01-16T09:59:40Z
Host: Linux 6.8.0-51-generic aarch64
CMake: 3.31.1
CMake generator: Unix Makefiles
CMake build tool: /bin/gmake
Configuration: Release
Algorithm Hint: ALGO_HINT_ACCURATE
CPU/HW features:
Baseline: NEON FP16
requested: DETECT
Dispatched code generation: NEON_DOTPROD NEON_FP16 NEON_BF16
requested: NEON_FP16 NEON_BF16 NEON_DOTPROD
NEON_DOTPROD (1 files): + NEON_DOTPROD
NEON_FP16 (2 files): + NEON_FP16
NEON_BF16 (0 files): + NEON_BF16
C/C++:
Built as dynamic libs?: NO
C++ standard: 11
C++ Compiler: /opt/rh/devtoolset-10/root/usr/bin/c++ (ver 10.2.1)
C++ flags (Release): -Wl,-strip-all -fsigned-char -W -Wall -Wreturn-type -Wnon-virtual-dtor -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -fvisibility-inlines-hidden -O3 -DNDEBUG -DNDEBUG
C++ flags (Debug): -Wl,-strip-all -fsigned-char -W -Wall -Wreturn-type -Wnon-virtual-dtor -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wundef -Winit-self -Wpointer-arith -Wshadow -Wsign-promo -Wuninitialized -Wsuggest-override -Wno-delete-non-virtual-dtor -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -fvisibility-inlines-hidden -g -O0 -DDEBUG -D_DEBUG
C Compiler: /opt/rh/devtoolset-10/root/usr/bin/cc
C flags (Release): -Wl,-strip-all -fsigned-char -W -Wall -Wreturn-type -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -O3 -DNDEBUG -DNDEBUG
C flags (Debug): -Wl,-strip-all -fsigned-char -W -Wall -Wreturn-type -Waddress -Wsequence-point -Wformat -Wformat-security -Wmissing-declarations -Wmissing-prototypes -Wstrict-prototypes -Wundef -Winit-self -Wpointer-arith -Wshadow -Wuninitialized -Wno-comment -Wimplicit-fallthrough=3 -Wno-strict-overflow -fdiagnostics-show-option -pthread -fomit-frame-pointer -ffunction-sections -fdata-sections -fvisibility=hidden -g -O0 -DDEBUG -D_DEBUG
Linker flags (Release): -L/ffmpeg_build/lib -Wl,–gc-sections -Wl,–as-needed -Wl,–no-undefined
Linker flags (Debug): -L/ffmpeg_build/lib -Wl,–gc-sections -Wl,–as-needed -Wl,–no-undefined
ccache: YES
Precompiled headers: NO
Extra dependencies: /lib64/libopenblas.so Qt5::Test Qt5::Concurrent /usr/local/lib/libpng.so /usr/lib64/libz.so Qt5::Core Qt5::Gui Qt5::Widgets Iconv::Iconv dl m pthread rt
3rdparty dependencies: libprotobuf ade ittnotify libjpeg-turbo libwebp libtiff libopenjp2 IlmImf tegra_hal
OpenCV modules:
To be built: aruco bgsegm bioinspired calib3d ccalib core cvv datasets dnn dnn_objdetect dnn_superres dpm face features2d flann fuzzy gapi hfs highgui img_hash imgcodecs imgproc intensity_transform line_descriptor mcc ml objdetect optflow phase_unwrapping photo plot python3 quality rapid reg rgbd saliency shape signal stereo stitching structured_light superres surface_matching text tracking video videoio videostab wechat_qrcode xfeatures2d ximgproc xobjdetect xphoto
Disabled: world
Disabled by dependency: -
Unavailable: alphamat cannops cudaarithm cudabgsegm cudacodec cudafeatures2d cudafilters cudaimgproc cudalegacy cudaobjdetect cudaoptflow cudastereo cudawarping cudev fastcv freetype hdf java julia matlab ovis python2 sfm ts viz
Applications: -
Documentation: NO
Non-free algorithms: NO
GUI: QT5
QT: YES (ver 5.15.16 )
QT OpenGL support: NO
GTK+: NO
VTK support: NO
Media I/O:
ZLib: /usr/lib64/libz.so (ver 1.2.7)
JPEG: build-libjpeg-turbo (ver 3.0.3-70)
SIMD Support Request: YES
SIMD Support: YES
WEBP: build (ver decoder: 0x0209, encoder: 0x020f, demux: 0x0107)
AVIF: NO
PNG: /usr/local/lib/libpng.so (ver 1.6.44)
TIFF: build (ver 42 - 4.6.0)
JPEG 2000: build (ver 2.5.0)
OpenEXR: build (ver 2.3.0)
GIF: NO
HDR: YES
SUNRASTER: YES
PXM: YES
PFM: YES
Video I/O:
FFMPEG: YES
avcodec: YES (59.37.100)
avformat: YES (59.27.100)
avutil: YES (57.28.100)
swscale: YES (6.7.100)
avresample: NO
GStreamer: NO
v4l/v4l2: YES (linux/videodev2.h)
Parallel framework: pthreads
Trace: YES (with Intel ITT)
Other third-party libraries:
Lapack: YES (/lib64/libopenblas.so)
Eigen: NO
Custom HAL: YES (carotene (ver 0.0.1))
Protobuf: build (3.19.1)
Flatbuffers: builtin/3rdparty (23.5.9)
OpenCL: YES (no extra features)
Include path: /io/opencv/3rdparty/include/opencl/1.2
Link libraries: Dynamic load
Python 3:
Interpreter: /opt/python/cp39-cp39/bin/python3.9 (ver 3.9.20)
Libraries: libpython3.9m.a (ver 3.9.20)
Limited API: YES (ver 0x03060000)
numpy: /home/ci/.local/lib/python3.9/site-packages/numpy/_core/include (ver 2.0.2)
install path: python/cv2/python-3
Python (for build): /opt/python/cp39-cp39/bin/python3.9
Java:
ant: NO
Java: NO
JNI: NO
Java wrappers: NO
Java tests: NO
Install to: /io/_skbuild/linux-aarch64-3.9/cmake-install
\n
\n
So install path is python/cv2/python-3. What makes me suspicious is that all the cuda modules are listed as unavailable.
I doubt that. Running strace -o file.txt python3 -c “import cv2; cv2.cuda.getCudaEnabledDeviceCount()” does not mention cudnn.
AastaLLL April 28, 2025, 7:15am 7
Hi,
Do you use Python3.9?
The package shared above is built for default Python 3.10.
Thanks.
No I’m using the default python 3.10.12. I’m running this in a docker container for which I provided the image.
I assume that:
Python 3:
Interpreter: /opt/python/cp39-cp39/bin/python3.9 (ver 3.9.20)
Libraries: libpython3.9m.a (ver 3.9.20)
Limited API: YES (ver 0x03060000)
numpy: /home/ci/.local/lib/python3.9/site-packages/numpy/_core/include (ver 2.0.2)
install path: python/cv2/python-3
specifies how opencv was built, neither the interpreter nor the numpy path exist in the container.
Hi, are there any updates?