How to install opencv-python for python3.6

Autonomous Machines Jetson & Embedded Systems Jetson Nano
#21

Oh yeah, forgot to mention the swapfile:

Almost the same sequence of commands from “air1kdf” above. just the first command I used was different:
sudo fallocate -l 4.0G /swapfile # this is the difference sudo chmod 600 /swapfile
sudo mkswap /swapfile sudo swapon /swapfile

Also, to make the swapfile activated during reboot, edit /etc/fstab:
$ sudo nano /etc/fstab

And add the following line:
/swapfile none swap 0 0

Save it and it should activate the swapfile after your next reboot.

#22

Although marked as “solved”, it really isn’t.
OpenCV4.1 IS available loaded with Jetpack 4.3. out of the box, but it does not appear as the “opencv-python” module, so it is not recognized as available by other packages such as stable-baselines. If you do a apt-install opencv you get version 3.2 and it appears as the module " python3-opencv" which is also not recognized by loading packages as OpenCV.
How can we get the pre-loaded OpenCV4.1 to appear as the requested module “opencv-python”?

#23

is there any real and working answer to this? i need opencv-contrib package to use object trackers.

edit: the answer is “build it yourself”. given script does not exist any more. this link will help you:

go to step 14 and follow the instructions. I didn’t use virtual environment. it will take around 3 hours.

#24

You may try the script by community member @mdegans: https://github.com/mdegans/nano_build_opencv

#25

Thanks @dkreutz

There are also pre-built docker images for JetPack 4.3 and 4.4 you can use as a base image:
https://hub.docker.com/r/mdegans/tegra-opencv/tags

That’ll save you a few hours of waiting for the script to build and is probably eaiser to test, deploy, and far more repeatable than the script itself. Make sure to use the --runtime nvidia flag to access the GPU or it’ll segfault when you try to “import cv2” or equivalent. Just:

sudo docker run --user $(id -u):$(cut -d: -f3 < <(getent group video)) --runtime nvidia -it --rm mdegans/tegra-opencv:latest

The --user bit is optional but recommended to avoid running the container as root.

And you can then:

 $ sudo docker run --user $(id -u):$(cut -d: -f3 < <(getent group video)) --runtime nvidia -it --rm mdegans/tegra-opencv:latest
[sudo] password for your_username_here: 
Unable to find image 'mdegans/tegra-opencv:latest' locally
latest: Pulling from mdegans/tegra-opencv
3b5e2c02f523: Already exists 
b9448035bb0a: Already exists 
ec9ea6732d09: Already exists 
515b0d7ffbf7: Already exists 
7663c88dd173: Already exists 
5dd4bd16e931: Already exists 
5231bd433c70: Already exists 
9e017cdad796: Already exists 
0c8ba4b0f40b: Already exists 
de2653fc4eda: Already exists 
21b3dc12ac76: Already exists 
18d3041d2711: Already exists 
8b4bed079038: Already exists 
efb313fc2a73: Already exists 
5c1e867b3d6d: Already exists 
aee12882d462: Pull complete 
c8c16c286602: Pull complete 
eacb1135637c: Pull complete 
92fdb5579a06: Pull complete 
4d54fa5df78f: Pull complete 
46d431fc04e0: Pull complete 
Digest: sha256:379826dbe07d5135615e02de0efc62da59ddb6097ca98e639f971db907b96776
Status: Downloaded newer image for mdegans/tegra-opencv:latest
I have no name!@e71c4e768a02:/usr/local/src/build_opencv$ python3
Python 3.6.9 (default, Apr 18 2020, 01:56:04) 
[GCC 8.4.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import cv2
>>> cv2.cuda.printCudaDeviceInfo(0) 
*** CUDA Device Query (Runtime API) version (CUDART static linking) *** 

Device count: 1

Device 0: "Xavier"
  CUDA Driver Version / Runtime Version          10.20 / 10.20
  CUDA Capability Major/Minor version number:    7.2
  Total amount of global memory:                 15823 MBytes (16591372288 bytes)
  GPU Clock Speed:                               1.38 GHz
  Max Texture Dimension Size (x,y,z)             1D=(131072), 2D=(131072,65536), 3D=(16384,16384,16384)
  Max Layered Texture Size (dim) x layers        1D=(32768) x 2048, 2D=(32768,32768) x 2048
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per block:           1024
  Maximum sizes of each dimension of a block:    1024 x 1024 x 64
  Maximum sizes of each dimension of a grid:     2147483647 x 65535 x 65535
  Maximum memory pitch:                          2147483647 bytes
  Texture alignment:                             512 bytes
  Concurrent copy and execution:                 Yes with 1 copy engine(s)
  Run time limit on kernels:                     No
  Integrated GPU sharing Host Memory:            Yes
  Support host page-locked memory mapping:       Yes
  Concurrent kernel execution:                   Yes
  Alignment requirement for Surfaces:            Yes
  Device has ECC support enabled:                No
  Device is using TCC driver mode:               No
  Device supports Unified Addressing (UVA):      Yes
  Device PCI Bus ID / PCI location ID:           0 / 0
  Compute Mode:
      Default (multiple host threads can use ::cudaSetDevice() with device simultaneously) 

deviceQuery, CUDA Driver = CUDART, CUDA Driver Version  = 10.20, CUDA Runtime Version = 10.20, NumDevs = 1

If you’re still on jetpack 4.3, you can use jp-r32.3.1-cv-4.3.0 instead of latest as a tag.

#26

Can you also import opencv_contrib packages like aruco?

#27