OpenCV4Tegra doesn't support GPU

Robotics & Edge Computing Jetson & Embedded Systems Jetson TX2
1

Hello, It seems that OpenCV4Tegra doesn’t support GPU.
I ran the same code with ros-kinetic-opencv3, I get the frame per second about 3 to 4.12
then, OpenCV4Tegra, I get the frame per second about 3 to 4.64

what is wrong?

my system:

NVIDIA Jetson TX2
L4T 28.2.1 [ 3.2.1 ]
Board: t186ref
Ubuntu 16.04.6 LTS
Kernel Version: 4.4.38-tegra
CUDA 9.0.252

Thank you for your help

2

Hi red8341h,

Opencv in jetpack has disabled gpu for long time. If you want opencv with gpu enabled, please download upstream opencv and install it with gpu enabled manually.

3

Thanks, WayneWWW,

I read an old post [url]https://www.jetsonhacks.com/2015/06/14/ros-opencv-and-opencv4tegra-on-the-nvidia-jetson-tk1/[/url], he mentioned this “you can use OpenCV4Tegra and get the best performance”

I am following this [url]https://www.jetsonhacks.com/2018/05/28/build-opencv-3-4-with-cuda-on-nvidia-jetson-tx2/[/url] but I am not sure if it works with me or not???

I get another one [url]https://jkjung-avt.github.io/opencv3-on-tx2/[/url]

but both, they didn’t add the opencv_contrib

4

Please try the installation script written by my colleague.

5

I’d like to try it but I also didn’t see the opencv_contrib in the script of your colleague.

all of needed, download it, then run the command:

./install_opencv3.4.0_TX2.sh

for JetsonTX2, right?

6

The new version has contrib enabled. Not sure if this matches your request.

https://github.com/AastaNV/JEP/blob/master/script/install_opencv4.0.0_Nano.sh

Actually, I think you could also reach opencv forum for more info.

7

Thank you so much, WayneWWW,

I saw opencv_contrib within it.

but can I use [url]https://github.com/AastaNV/JEP/blob/master/script/install_opencv4.0.0_Nano.sh[/url] for JetsonTX2 without any problem? if YES?

I will use it immediately

8

I downloaded it, then I modified version https://github.com/AastaNV/JEP/blob/master/script/install_opencv4.0.0_Xavier.sh from 4.0.0 to 3.4.0 , also CUDA_ARCH_BIN=“7.2” to CUDA_ARCH_BIN=“6.2”

finally, I tried to run it:

./install_opencv3.4.0_TX2.sh

I get this output:

Usage: ./install_opencv3.4.0_TX2.sh <Install Folder>

what is wrong?

9

I get it [url]https://github.com/AastaNV/JEP/issues/15[/url]

10

I built OpenCV from source but I didn’t note any difference between ros-kinetic-opencv3 and opncv with gpu
the same frame per second about 3- 4.86

what is wrong???

python
Python 2.7.12 (default, Oct  8 2019, 14:14:10) 
[GCC 5.4.0 20160609] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import cv2
>>> cv2.__version__
'3.4.1-dev'
>>>

sudo ./tegrastats 
RAM 1238/7846MB (lfb 1339x4MB) CPU [0%@345,off,off,0%@345,1%@345,1%@345] EMC_FREQ 15%@204 GR3D_FREQ 0%@140 APE 150 BCPU@45.5C MCPU@45.5C GPU@44C PLL@45.5C Tboard@40C Tdiode@43C PMIC@100C thermal@44.9C VDD_IN 1757/1816 VDD_CPU 229/202 VDD_GPU 152/152 VDD_SOC 381/392 VDD_WIFI 57/78 VDD_DDR 443/441

any ideas or suggestions?

11

Replied your question in

[url]https://devtalk.nvidia.com/default/topic/1065010/jetson-tx2/does-opencv_dnn-use-gpu-/post/5393316/#5393316[/url]