Deepstream peoplenet transformer error in src_bin_muxer

2000520310002 · May 10, 2024, 10:46am

Please provide complete information as applicable to your setup.

• GPU
• DeepStream Version - 7.0
• None
• A100
• Bugs
• src_bin_muxer/gstnvstreammux.cpp

Error:
max_fps_dur 8.33333e+06 min_fps_dur 2e+08
** ERROR: main:706: Failed to set pipeline to PAUSED
Quitting
ERROR from src_bin_muxer: Batch size not set
Debug info: gstnvstreammux.cpp(1523): gst_nvstreammux_change_state (): /GstPipeline:pipeline/GstBin:multi_src_bin/GstNvStreamMux:src_bin_muxer
App run failed

fanzh · May 11, 2024, 6:11am

which sample are you testing? from the error log, nvstreammux’s batch-size is not set.

2000520310002 · May 13, 2024, 6:31am

I was trying to run the peoplenet transformer
How to set nvstreammux’s batch-size?

fanzh · May 13, 2024, 6:33am

for example, g_object_set (G_OBJECT (streammux), “batch-size”, 1, NULL);

2000520310002 · May 13, 2024, 7:26am

github.com

NVIDIA-AI-IOT/deepstream_tao_apps/blob/master/models/peoplenet_transformer/config.pbtxt

################################################################################
# Copyright (c) 2023 NVIDIA Corporation.  All rights reserved.
#
# Permission is hereby granted, free of charge, to any person obtaining a
# copy of this software and associated documentation files (the "Software"),
# to deal in the Software without restriction, including without limitation
# the rights to use, copy, modify, merge, publish, distribute, sublicense,
# and/or sell copies of the Software, and to permit persons to whom the
# Software is furnished to do so, subject to the following conditions:
#
# The above copyright notice and this permission notice shall be included in
# all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
# THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
# FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
# DEALINGS IN THE SOFTWARE.

This file has been truncated. show original

I have a config file and a label file
where can I edit batch size?
Also, max batch size is already defined in config file here

github.com

NVIDIA-AI-IOT/deepstream_tao_apps/blob/master/models/peoplenet_transformer/config.pbtxt#L25


      
          # IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
          # FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
          # THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
          # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
          # FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
          # DEALINGS IN THE SOFTWARE.
          ################################################################################
          
          name: "peoplenet_transformer"
          platform: "tensorrt_plan"
          max_batch_size: 1
          default_model_filename: "resnet50_peoplenet_transformer_op17.onnx_b1_gpu0_fp16.engine"
          input [
            {
              name: "inputs"
              data_type: TYPE_FP32
              dims: [ 3, 544, 960]
            }
          ]
          output [
            {

fanzh · May 13, 2024, 7:49am

Did you modify the code? without the modification, I can’t reproduce this issue. here is my log. log.txt (2.8 KB)
peoplenet_transformer/config.pbtxt is model configuration for triton mode. it is not related to “Batch size not set” error.

2000520310002 · May 13, 2024, 8:15am

No
I started a fresh container for deepstream-7.0
as stated on quickstart guide on website:
docker run --gpus "device=0" -it --rm -v /tmp/.X11-unix:/tmp/.X11-unix -e DISPLAY=$DISPLAY -e CUDA_CACHE_DISABLE=0 nvcr.io/nvidia/deepstream:7.0-triton-multiarch
went to this path
/opt/nvidia/deepstream/deepstream-7.0/samples/triton_tao_model_repo/peoplenet_transformer
and ran the command as:
deepstream-app -c config.pbtxt

initially it was giving an error on “output width not set”
so set this environment variable as export USE_NEW_NVSTREAMMUX=yes
as listed here: Gst-nvstreammux New — DeepStream documentation 6.4 documentation
Then after running the same config.pbtxt again I got batch size error

Can you tell me how to run peoplenet_tranformer?

fanzh · May 13, 2024, 8:34am

this command-line is wrong becasue config.pbtxt is not deepstram-app’s configuration file. if you want to test . you can use deepstream-app -c source1_primary_detector_peoplenet_transformer.txt. this file is in opt\nvidia\deepstream\deepstream-7.0\samples\configs\deepstream-app-triton. please refer to opt\nvidia\deepstream\deepstream-7.0\samples\configs\deepstream-app-triton\README.

2000520310002 · May 15, 2024, 4:17pm

actually I am getting same batch size error on all apps, it is not specific to peoplenet_transformer
for example

2000520310002 · May 15, 2024, 4:54pm

ran it on 2 different gpus RTX4500 , A100
same error is showing in new container

output width not step → solved by setting environment variable export USE_NEW_NVSTREAMMUX=yes
Batch size not set
same error of batch size on both gpus
on any app, not just peoplenet

fanzh · May 16, 2024, 9:26am

config_infer_primary.txt is the configuration file of nvinfer. if you wan’t test deepstream-app, please refer to my comment on May 13.

2000520310002 · May 18, 2024, 5:58am

Thank you for your response.
When following the peoplenet_transformer app instructions from readme I am getting the following error on running the script prepare_ds_triton_tao_model_repo.sh in samples directory

fanzh · May 19, 2024, 3:23pm

Thanks for the sharing! currently you can download the newer onnx model, here is the converting command-line.

2000520310002 · May 21, 2024, 7:31am

Thanks for the commands.
I am getting this error

My engine file and onnx file are here

Is it a path issue?
Or I am doing something wrong?

Thanks

fanzh · May 21, 2024, 7:51am

the path and batch-size of engine need to be updated. please copy this patch pt_onnx.sh (5.1 KB) to /opt/nvidia/deepstream/deepstream/samples, then run this patch to generate an engine. after finishing, there will be an engine model.plan in /opt/nvidia/deepstream/deepstream/samples/triton_tao_model_repo/peoplenet_transformer/1. then try deepstream-app again.

2000520310002 · May 21, 2024, 5:53pm

I generated the model.plan at the path mentioned above.

where to update the path and batch size of engine?

fanzh · May 22, 2024, 5:25am

if you have run the pt.onnx_sh, please share the result of “ll /opt/nvidia/deepstream/deepstream/samples/triton_tao_model_repo/p
eoplenet_transformer/1/” ? pt.onnx_sh will create the engine model.plan.
if still can’t work. could you share more logs by the following step?

export GST_DEBUG=6 && deepstream-app -c source1_primary_detector_peoplenet_transformer.txt  >1.log 2>1.log

2000520310002 · May 23, 2024, 7:07am

1.log (4.7 MB)
Seems, model.plan was plan was correctly generated still the app is not running

fanzh · May 23, 2024, 7:35am

it is because creating render plugin failed.

you can use filesink to make a recording. first please run /opt/nvidia/deepstream/deepstream/user_additional_install.sh to install software encoder because A100 does not support hardware encoding. then please set type=1 in [sink0], and set enable=1, enc-type=1 in [sink1].
please refer to this topic. A100 is not display card. you can set up a virtual display.

2000520310002 · May 27, 2024, 7:53pm

Hi
I am still getting this error

1.log (10.1 MB)

Topic		Replies	Views
Nvvideoconvert_issue DeepStream SDK	12	1639	February 28, 2022
Deepstream error DeepStream SDK	6	314	February 9, 2024
Faces and bags not detected on peoplenet DeepStream SDK	6	177	May 31, 2024
DeepStream SDK 5.0 with Nvidia GTX 960(4GB RAM) - deepstream-app throwing Bus error DeepStream SDK	9	734	October 12, 2021
Runtime errors when running the human pose estimation application DeepStream SDK tensorrt , cuda , ubuntu	25	4789	October 12, 2021
Problem facing with DeepStream App DeepStream SDK	13	2387	October 12, 2021
How to fix in testing peoplenet? ERROR: <gst_nvinfer_parse_config_file:1242>: failed ERROR from primary_gie: Configuration file parsing failed DeepStream SDK	12	1607	October 12, 2021
Issues with running inference on multiple rtsp streams in deepstream-imagedata-multistream DeepStream SDK jetson-inference	24	616	August 7, 2024
Deepstream-rtsp-in-rtsp-out : AttributeError: 'NoneType' object has no attribute 'set_property' DeepStream SDK jetson , deepstream	22	72	October 9, 2024
Unable to run PeopleNet - Engine file generation failure DeepStream SDK deepstream	3	12	January 16, 2025

Deepstream peoplenet transformer error in src_bin_muxer

Related topics