**• |
Hardware Platform--------Jetson** |
**• |
DeepStream Version-------5.0** |
**• |
JetPack Version -----------4.4** |
**• |
TensorRT Version----------7.1.3.0** |
I suppose that deepstream-app formulates a gstreamer pipeline and then just runs it. Please correct me if I am wrong.
I wonder if I can get the formulated pipeline. Because I need to add a gstreamer ‘videoflip’ element. My camera is mounted upside-down {unfortunately}.
Thanks :)
Hi, you are right. That is actually the way I generated the pipelines I use for DeepStream 4.0:
https://developer.ridgerun.com/wiki/index.php?title=DeepStream_pipelines
I haven’t documented pipelines dos DeepStream 5.0 but they are almost the same. Here is how I did it:
- Add the flag to generate a GStreamer dot file to the command to run a deepstream example:
GST_DEBUG_DUMP_DOT_DIR=.
- Generate a PNG from the dot file:
dot -Tpng <name>.dot > pipe.png
- Then check the dot file for the elements and properties and add them to your pipeline
1 Like
Thanks @miguel.taylor
Sorry if my question is naive. Are you indicating to modify .../deepstream_sdk_5.0_x86_64/opt/nvidia/deepstream/deepstream/sources/apps/sample_apps/deepstream-app/deepstream_app_main.c
?
If yes, then where should I add the flag in the file?
Hi,
Sorry if it wasn’t clear. The flag is added before running the application that uses GStreamer. For example:
GST_DEBUG_DUMP_DOT_DIR=. gst-launch-1.0 ...
GST_DEBUG_DUMP_DOT_DIR=. ./deepstream-test1 ...
GST_DEBUG_DUMP_DOT_DIR=. deepstream-app ...
1 Like