TX2 with DeepStream for MP4v2 output

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hi all:

I want to use TX2 with deepstream to generate mp4v2 file inculde h264 and aac for streaming.

I want to know on TX2, do we have example to demo how to use deepstream to speedup it with TX2 powerful HW?

thank you

2

Hi,
You may refer to deepstream-app. We have sample config file for TX2.

$ deepstream-app -c source12_1080p_dec_infer-resnet_tracker_tiled_display_fp16_tx2.txt

You can configure sink to ‘Encode + File Save’
https://docs.nvidia.com/metropolis/deepstream/dev-guide/index.html#page/DeepStream_Development_Guide%2Fdeepstream_app_config.3.2.html%23wwpID0ENHA

And integrate audio part into deepstream-app. Source code is in

deepstream_sdk_v4.0.1_jetson\sources\apps\sample_apps\deepstream-app

For audio/video encoding into mp4, you may refer to
https://devtalk.nvidia.com/default/topic/1064809/jetson-nano/how-to-add-empty-audio-track-to-video-stream/post/5392060/#5392060

3

Hi,

my video frame is the result after my analytics with boundbox output, and both video and audio original source is from rtsp streaming, on this case, how to use deepstream to generate a rtsp streaming to output the reslut with HW benefit? I could use ffmpeg to do this, but the performance is poor.

thank you

4

I could build deepstream-app but I don’t know how to set the video and audio as input to leverage deepstream benfit.

5

Hi,
On Jetson platforms, there is not hardware audio encoder/decoder and you have to use native gstreamer plugin such as avdec_aac
https://gstreamer.freedesktop.org/documentation/libav/avdec_aac.html?gi-language=c

Also deepstream-app is video only and you may integrate audio decoding into it.

Below is a sample pipeline of video playback. FYR.

$ gst-launch-1.0 filesrc location=videoplayback1.mp4 ! qtdemux name=d \
  d.video_0 ! queue ! h264parse ! nvv4l2decoder ! nvoverlaysink \
  d.audio_0 ! queue ! avdec_aac ! audioconvert ! alsasink