Skip to content

Latest commit

 

History

History
157 lines (116 loc) · 6.47 KB

TESTING.md

File metadata and controls

157 lines (116 loc) · 6.47 KB
Raw

Verification Testing for OpenCV Installation

As OpenCV is a complex beast, to ensure the full installation of OpenCV is working correctly we perform the following tests.

All tested with OpenCV 3.x / 4.x and Python 3.x.

Assumes that git and wget tools are available on the command line or that similar tools are available to access git / download files.

(On-site at Durham University: For testing on MS Windows, download and unzip example files from source URL as specified - here - onto the user space J:/ drive. On Linux run the associated opencv ... .init shell script before testing.).

Test #1 - check versions:

(for testing on MS Windows at Durham (DUDE system), open all .py scripts from J:/ in IDLE and run module from there)

git clone https://github.com/tobybreckon/python-examples-ip.git
cd python-examples-ip
python3 ./version.py

Result #1:

  • Text output to console such that:
We are using OpenCV: CCC
.. do we have the OpenCV Contrib Modules: True
.. do we have the OpenCV Non-free algorithms: True
.. do we have the Intel Performance Primitives (IPP): 
.. version: <???> (in use: True)
We are using numpy: <???>
We are using matplotlib: <???>
.. and this is in Python: PPP ??? (64 bit)

Check Video I/O (OS identifier: MMM)
... available camera backends:  LLL
... available stream backends:  LLL
... available video writer backends: LLL

Available Cuda Information:
... ['NVIDIA CUDA: YES (ver NNN, RRR)', 'NVIDIA GPU arch: ???', 'NVIDIA PTX archs: ZZZ']

GGG

DNN module CUDA backend/target availability : 
... DNN_TARGET_CUDA:            True
... DNN_TARGET_CUDA_FP16:       True
... DNN_TARGET_CPU:             True
... DNN_TARGET_OPENCL:          True
... DNN_TARGET_OPENCL_FP16:     True

OpenCL available (within OpenCV) ? : True

Available CPU Optimizations (*: build enabled; ?: not CPU supported):
... ???
  • such that CCC >= 4.3.x (or higher), PPP > 3.x, MMM is sensible for the OS in use, each of the LLL list are sensible (may not all be identical) and ideally include FFMPEG + GSTREAMER in addition to V4L/V4L (for MMM = linux*), QT (for MMM = darwin) or DSHOW / MSMF (for MMM = win*), NNN > 10.x, ZZZ includes cuDNN: Yes, GGG is sensible if a CUDA compatible GPU is present + configured and ??? = (doesn't matter). In addition, for maximum performance RRR ideally includes CUFFT CUBLAS FAST_MATH.

[ N.B. to build with Non-free algorithms set OPENCV_ENABLE_NONFREE=TRUE in CMake ]

Test #1a - check we have all the correct functionality:

Only to be used if building OpenCV from source (any platform) or using a pre-built version from a 3rd party repository (on Linux / Mac OS).

.. (as per test 1 for steps 1 + 2)
python3 ./checkbuildinfo.py

Results #1a

  • The resulting build element information should ideally contain the following as a subset of the configuration (where present indicates a reference to a system file or a version number that tells us it is present on the system / in the build):
build element Linux MS Windows
Platform: ... Configuration: Release Release
Built as dynamic libs: YES YES
OpenCV modules: contains: core flann ... imgproc ... ml ... imgcodecs ... videoio .. xfeatures2d ... dnn ... dnn_objdetect ... ximgproc ... optflow ... stitching ... cuda <-- same as Linux
Non-free algorithms: YES YES
QT: NO NO
GTK+ 2.x: YES NO
ZLib: present present
JPEG: present present
WEBP: present present
PNG: present present
JPEG 2000: present present
OpenEXR: present present
PNG: present present
FFMPEG: YES (all sub-elements) NO
GStreamer: YES (all sub-elements) NO
V4L/V4L2: present NO
XINE: YES NO
gPhoto2: YES NO
Parallel framework: pthreads ???
Use IPP: present NO
NVIDIA CUDA: YES ( >= 10.x) NO
OpenCL: YES YES
  • elements in bold required for OpenCV 4.3.x and later on Linux (to enable GPU use for CNN/DNN inference via DNN module).

Test #2 - check image read + window display functions:

(for MS Windows download image from here and save as example.jpg in directory python-examples-ip-master)

.. (as per test 1 for steps 1 + 2)
wget https://upload.wikimedia.org/wikipedia/commons/b/b4/JPEG_example_JPG_RIP_100.jpg
mv JPEG_example_JPG_RIP_100.jpg example.jpg
python3 ./smooth_image.py

Result #2:

  • image is displayed in window (a blurred version of example.jpg), pressing 'x' closes window.

Test #3 - test camera access and video write function:

.. (as per test 1 for steps 1 + 2)
<<< connect a usb web-cam (any web-cam)
python3 ./save_video.py -c 0

Result #3:

  • possible error message such as "???... ???? Unable to open source 'input.avi'" - ignore this.
  • possible error message relating to inability to read from camera or video device (try with -c 1 or greater to address different video devices)
  • a window appears with live video from camera, pressing 'x' closes window and ends program (saving a video file).
  • video file saved as output.avi file can be played in vlc or similar.
  • Ignore "cannot open video messages" from ffpmeg/xine or similar (this is it failing to open video file as alt. to camera, not write from camera).

Test #4 - test video read function:

(for MS Windows download video from here and save as video.avi in directory python-examples-ip-master)

.. (as per test 1 for steps 1 + 2)
wget https://download.blender.org/peach/bigbuckbunny_movies/big_buck_bunny_720p_surround.avi
mv big_buck_bunny_720p_surround.avi video.avi
python3 ./capture_video.py

Result #4: