Broken frame image returned by get_nvds_buf_surface

• Hardware Platform (Jetson / GPU)
Jetson Xavier AGX
• DeepStream Version
DeepStream 6.0
• JetPack Version (valid for Jetson only)
JetPack 4.6
• Issue Type( questions, new requirements, bugs)
question
• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)

The sample application is included. Substitute your rtsp or file uri in self.add_source() call on line 313 and run the script. The script writes frames into source_0.jpg file.

• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)

Python 3.x and OpenCV are required (I used Python 3.6.9 and OpenCV version 4.5.4, but I guess that other versions are acceptable).

I’m running a pipeline like the following:

gst-launch-1.0 -e \
        nvstreammux name=mux width=960 height=540 batch-size=1 live-source=1 \
        ! nvstreamdemux name=demux \
	uridecodebin uri=file:/home/zabra/workspace/streams/1.mp4 \
        ! videorate \
        ! nvvideoconvert ! "video/x-raw(memory:NVMM),format=RGBA,framerate=2/1" \
        ! queue \
        ! mux.sink_0 \
        demux.src_0 \
        ! queue \
        ! nvdsosd \
        ! nvegltransform \
        ! nveglglessink \

This is a test pipeline, the actual pipeline contains nvinfer and nvtracker between nvstreammux and nvstreamdemux.

I’m trying to get frame image using a buffer probe on nvdsosd src pad. However, when I decode the data returned by get_nvds_buf_surface() call, I see the broken image (just garbage on screen, black image etc…). Meanwhile nveglglessink shows correct frames and the pipeline works fine.

When I put an inference between the muxer and demuxer, I get broken images until the network detects a first object on video. As soon as a first object is detected, the frame images become normal.

I suppose I’m doing something wrong with the pipeline. Could you please help me to resolve the issue?
test_pipeline.py (11.7 KB)

Your code is wrong. The pipeline is not build successfully. Please refer to deepstream_python_apps/apps at master · NVIDIA-AI-IOT/deepstream_python_apps · GitHub for constructing pipeline with gst-python and deepstream.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.