python: added detectnet-video.py and imagenet-video.py

[aconchillo]

Added two new examples that add support for using detectnet and imagenet on
video files.

[aconchillo]

This PR depends on dusty-nv/jetson-utils#10

[aconchillo]

updated the two example so closing the renderer window (if --render) also stops the example.

[dusty-nv]

Thanks @aconchillo , appreciate it. I will merge these and your PR's for jetson-utils as soon as I finish making a few other unrelated modifications to the repo.

[aconchillo]

@dusty-nv absolutely. i have also updated the jetson-utils PR to use mapped memory instead. i assume it's hopefully the same in terms of performance. the reason to not copy the frame to GPU-only memory is because of jetson.utils.saveImageRGBA only works with mapped memory. see aconchillo/jetson-utils@a557a85#commitcomment-34771648

[maponarooo]

An error occurs at cmake ../
What is the problem?

jetson@jetson-nano:~/jetson/acon/jetson-inference/build$ cmake ../

Note: this project uses git submodules in the source tree.
if you haven't already, run the following command from
the project's root directory:

       git submodule update --init

-- using patched FindCUDA.cmake
-- CUDA version: 10.0
-- CUDA 10 detected, enabling SM_72
CMake Error at CMakeLists.txt:74 (find_package):
Could not find a configuration file for package "OpenCV" that is compatible
with requested version "3.0.0".

The following configuration files were considered but not accepted:

/usr/lib/aarch64-linux-gnu/cmake/opencv4/OpenCVConfig.cmake, version: 4.1.1

-- Configuring incomplete, errors occurred!

[dusty-nv]

I wonder if your repo is out-of-date, because this issue was fixed for JetPack 4.3 here:

jetson-inference/CMakeLists.txt

Line 74 in 8846dcb

find_package(OpenCV COMPONENTS core calib3d)

[maponarooo]

I wonder if your repo is out-of-date, because this issue was fixed for JetPack 4.3 here:

jetson-inference/CMakeLists.txt

Line 74 in 8846dcb

find_package(OpenCV COMPONENTS core calib3d)

I got it~
Thank you!

[fauziacharuna]

hi @aconchillo sorry for my bad english, i'm very beginner in jetson inference. I was completely clone your forks repo which contain detecnet-video, but when im running in my jetson inference, error saying 'module' object has no attribute 'VideoSource'. How to solve this problem? is that the opencv must version 3? or rebuild the jetson inference? thanks for your advice.

[bl6594]

Not sure why you chose to use the aconchillo's version instead of main repo. I am guessing you want to use gstreamer. As to the cv version, I am using 4.1.1 and it seems working for me. BTW, I tried aconchillo's version before, and found that there was a memory leak at that time, then switched back to the main repo.

[aconchillo]

I haven't played with this for a long time. I will try to revive it and get it up to date. @bl6594 does the main repo support detection on video files? if so there might not be any need to update this.

[dusty-nv]

@aconchillo the jetson-inference dev branch now supports video files, RTP/RTSP streams, ect from both C++ and Python. So you may not need to update it. I will be merging it into master as soon as I am done with documentation. For now, see here:

https://github.com/dusty-nv/jetson-inference/blob/dev/docs/aux-streaming.md

[aconchillo]

@dusty-nv awesome! thanks!