Three source is same and Synchronize，just one source have object; WHY?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

gst-launch-1.0 rtspsrc location=rtsp://gelanbo-rtsp:8556/test ! rtph264depay ! nvv4l2decoder ! m.sink_0
rtspsrc location=rtsp://gelanbo-rtsp:8556/test ! rtph264depay ! nvv4l2decoder ! m.sink_1
rtspsrc location=rtsp://gelanbo-rtsp:8556/test ! rtph264depay ! nvv4l2decoder ! m.sink_2
nvstreammux name=m batch-size=1 width=1920 height=1080 ! nvinfer config-file-path=pgie_config.txt ! nvvideoconvert ! nvdsosd ! nvmultistreamtiler rows=3 columns=3 width=1920 height=1080 ! nvvideoconvert ! nvv4l2h264enc ! h264parse ! flvmux ! rtmpsink location=rtmp://192.168.3.201:1935/live/tiler sync=false

When I use this command to execute, there is one and only one small window with a detected object. After many experiments, the source of the detected object is not sure which way, I don’t understand.

• Hardware Platform (Jetson / GPU) xavier and xavier nx
• DeepStream Version5.0
• JetPack Version (valid for Jetson only)
• TensorRT Version
• NVIDIA GPU Driver Version (valid for GPU only)
• Sample application and the configuration file content
• Reproduce steps
• Reproducing rate and duration

3

You need to use nvmultistreamtiler before nvdsosd.

4

I use this pipeline with yolov3 model, it works well.I think onnx model or other point

5

nvmultistreamtiler should be put before nvdsosd. And you have 3 sources, it is better to set nvstreammux batch-size as 3.

6

nvmultistreamtiler before nvdsosd ；The above phenomenon will still occur 。

7

If the pipeline works with other models, you may need to check the model you are using.

8

Ok I want know why nvmultistreamtiler before nvdsosd?

9

Because nvosd will change something nvmultistreamtiler needs.