Missing events when vehicle cross the line

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

• Hardware Platform (Jetson )
• DeepStream Version 6.0
• BOARD: t210ref
I am using yolov7 tiny model for vehicle detection and I am using nvdaanlytics for line crossing .
so the main issue is some time vehicle cross the line but it missed the event.

Currently i am using the model with minimum threshold value to get all the possible events so due to this tracker id also got changed.
But it is not the concern , the issue is missing event ,the vehicle cross the line but it didn’t generate the events.

I have also tried to add two more line to check if one is missed then other will capture but still few events are missing .
Kindly guide me how to resolve this missing event issue.
I have attached the pipeline.
config_pipeline.txt (4.8 KB)
nvdaanalytics_config.txt (1.8 KB)

3

please refer to this topic. maybe it is not the same issue. here are some questions:

  1. which sample are you testing？ did you modify the code?
  2. could you share the input video? and Please point out the not-working timespan.
4

Hi Fanzh ,
Yes I have modify the code for smart records.
I am using live video , so I don’t have the video at this time .
I am suspecting that the frames are dropping like initially the object is detected but during the line crossing the frame is dropping and the vehicle is missed.
can you help me to verify that ?

5

tracker can recognize the object again even some frames are dropped. as you know, if too many frames are dropped, tracker will give the object a new id. this will have an affect to line crossing results.

6

So how to resolve this too many frame drop issue ?
How I can analyze the app performance , latency?

7

can you make a video recording to show the “mssing events” issue? do you mean receiving data from live video has too many frame drop issue? it should be a network issue.

please refer to topic for performance analysis.
please refer to topic for fps checking.
please refer to topic for latency checking.
please use nvidia-smi to check GPU usage.

8

I have done the analysis.
there is no latency issue.
I am using rtsp camera which provide 10fps stream.
Now I have three problems->
1 - The frame some time corrupting when object moves.
2 - Gpu utilization is too high, initially when I launched the app fps was 10
but it is dropped to 5 fps after couple of hours.
3 - In console some time print the below error->
reference in DPB was never decoded
Please Help me to resolve the errors.

9

it is the same with question 3. decoding faild caused mosaic. maybe there are two cases. one is packets loss caused decoding failed. the other one is decoder bug. about case 1, please fix the the network issue. about case 2, please dump some video by this command “gst-launch-1.0 rtspsrc location=rtsp://xx ! rtph264depay ! h264parse ! mux. mpegtsmux name=mux ! filesink location=output.ts”
.

please make sure the source’s fps is stable. you can check it by this command “gst-launch-1.0 rtspsrc location=rtsp://xxx ! rtph264depay ! h264parse ! queue ! nvv4l2decoder ! fpsdisplaysink”。