PipeTuner

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

**• Hardware Platform- Jetson
**• DeepStream Version- 6.2
**• JetPack Version (valid for Jetson only)-5.1
**• TensorRT Version- 8.5.2.2

hello,
I am encountering an issue while using the Pipetuner tool with our custom model, configuration files, and videos. Here’s a detailed description of the problem:
I have converted videos in .mp4 format and according to video resolution made changes in the Pipetuner container and committed these changes.
I referred this forum: PipeTuner Custom Resolution Support

During the execution of the launch.sh file, I am encountering the following issue:
2024-06-10 06:35:41,054 root INFO send backend init
2024-06-10 06:35:41,054 root INFO creating optimizers…
2024-06-10 06:35:41,054 root INFO done. created 2

  • Serving Flask app ‘ds_bbo_frontend_server’
  • Debug mode: on
    2024-06-10 06:35:41,068 root INFO init jobs done
    2024-06-10 06:35:41,084 root INFO progress: 0% (0/40)
    Launching server on: http://0.0.0.0:50733
    received /init call
    2024-06-10 06:35:41,092 root INFO wait backend ready
    [server core] initializing
    2024-06-10 06:35:42,093 root INFO wait backend ready
    2024-06-10 06:35:44,397 root WARNING DsApp score is 0. number of 0 scores:1
    2024-06-10 06:35:44,404 root WARNING DsApp score is 0. number of 0 scores:2
    2024-06-10 06:35:46,209 root WARNING DsApp score is 0. number of 0 scores:3
    2024-06-10 06:35:46,210 root WARNING DsApp score is 0. number of 0 scores:4
    2024-06-10 06:35:48,333 root WARNING DsApp score is 0. number of 0 scores:5
    2024-06-10 06:35:48,333 root WARNING DsApp score is 0. number of 0 scores:6
    2024-06-10 06:35:48,635 root ERROR Too many 0 scores from DS app. We stop all processes. Please check DS app logs
    2024-06-10 06:35:48,635 root ERROR OPTIMIZATION not completed!
    [worker 1] up and running!
    [result sender] up and running!
    [worker 0] up and running!
    received /reset call
    [server core] resetting
    stopping result senders…
    done.
    stopping workers…
    number of workers to stop: 2
    number of workers to stop: 0
    done.

!!! Press CTRL+C key to end PipeTuner !!!

I am looking for guidance on what might be causing the DsApp score to be 0 and how to resolve it. Could you also tell me what is dsapp score?
Any suggestions would be greatly appreciated.

Thank you!

3

Does the original sample configuration work?

4

Hi, yes, the default configurations work fine.
What exactly is DsApp score? And, why am I getting it as 0?

5

What and how did you configure the PipeTuner?

6

I am sharing the PipeTuner config files I have used for custom model.
The resolution of the videos are 3840 x 2160.
files.zip (4.2 KB)

7

Have you tried the “config_infer_primary_yoloV5_320.txt” with deepstream-app? Can the bboxes be detected with your configuration?

8

Yes. I have tried, and it’s working.

9

Can you check the log files under the pipe-tuner-sample/output/ directory first?

E.G. I can find the “pipe-tuner-sample/output/SDG_sample_PeopleNet-ResNet34_NvDCF-ResNet50_MOT.yml_output/checkpoints/DsAppRun_output_20240611_010614/log_DsAppRun.log” after running the sample.

10

The log file is not generated when I’m running our custom model and videos in Pipetuner tool.

11

No log after the PipeTuner stopped? Even the server and client logs are not in the directory?

12

client and server logs are generated in directory, but log_DsAppRun.log does not get generated.

13

Can you upload the logs for checking?

14

Sure, I am sharing the client and server log files.
log_.zip (1.4 KB)

15

Hello

Not intending to hijack this thread but I am also experiencing similar issues trying to run YoloV7 with PipeTuner.
I am running on:

  • Hardware Platform: GeForce RTX 4090 Laptop GPU
  • Deepstream Version - 7.0

With one configuration I received the Too many 0 scores from DS app. We stop all processes. Please check DS app logs, With a similar configuration I end up with the logs attached where the ETA increases indefinitely after the first DsApp score message.

I can confirm this model, pgie config, and tracker config does show bounding boxes in the application its used in.
I can also run the sample provided by PipeTuner and it works as expected.

libnvdsinfer_custom_impl_Yolo.so used in my pgie config was copied from my applications DeepStream 7.0 docker container.

Thanks
log_client_2024-06-17_10-33-37.txt (4.5 KB)
log_DsAppRun.log (3.4 KB)
yolov7-pgie-config.txt (850 Bytes)

16

file.zip (3.2 KB)

Thank you, michaells, for sharing files. I will try with Yolo V7. I have a question. Previously, when I tried with my custom config files and videos, I got .csv file in output in that there is a score column, which has a 0. Could you please tell me what that is??
I am sharing the .csv file.

17

Hi Riya

Just to clarify I am experiencing the same issue as you but let me know if you get YoloV7 working for you!

18

Hi Fiona

I’m wondering if you could provide any guidance on how I can start to debug this given the logs.

Thanks

19

Hi PipeTuner user,

(1) Make sure the YOLO library is correctly mapped into DS container, and PGIE custom-lib-path points to the correct path in DS container.
(2) Make sure all the model related file paths are correct in PGIE config, like onnx-file, model-engine-file
(3) Once they launch PipeTuner, a DS container is launched. Try running DS app inside that DS container and see if the output is correct.

20

@michaells and @riya_shastri

Can you try our suggestions?

21

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks