Please provide complete information as applicable to your setup.
• Hardware Platform (Jetson / GPU) Tesla T4
• DeepStream Version 6.0
• TensorRT Version 8.0.0.1
• NVIDIA GPU Driver Version (valid for GPU only) 470.63.01
I’m trying to use nvblender
to overlay segmentation masks onto the original video passed into a TensorRT model. First of all, is this an appropriate use case?
Second of all, my program segfaults every time I pass video into it. The portion of the pipeline creating the problem is the nvblender
:
queue ! nvsegvisual !
/ \
... ! nvinfer ! tee ! nvblender ! ...
\ /
queue !
If I only give the blender one video stream, it doesn’t crash. The problem is isolated to this portion of the pipeline, as I can recreate the problem on the command line:
GST_DEBUG=5 sudo gst-launch-1.0 videotestsrc ! tee name=t ! queue ! nvblender name=b t. ! queue ! b.sink_1
Setting pipeline to PAUSED ...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
New clock: GstSystemClock
Redistribute latency...
Caught SIGSEGV
#0 0x00007f2aeddbbbb9 in __GI___poll (fds=0x5578e816ff90, nfds=2, timeout=-1)
#1 0x00007f2aee3036e9 in () at /usr/lib/x86_64-linux-gnu/libglib-2.0.so.0
#2 0x00007f2aee303a82 in g_main_loop_run ()
#3 0x00007f2aee8679f3 in gst_bus_poll ()
#4 0x00005578e506abb3 in ()
#5 0x00005578e5069cb7 in ()
#6 0x00007f2aedcc8c87 in __libc_start_main (main=
#7 0x00005578e506a0da in ()
Spinning. Please run 'gdb gst-launch-1.0 8874' to continue debugging, Ctrl-C to quit, or Ctrl-\ to dump core.
If this isn’t how I’m supposed to add the segmentation mask onto the original video, how should I do that?