How to increase number of file sources for DeepStream 3D Action Recognition App

Please provide complete information as applicable to your setup.

**• Hardware Platform (Jetson / GPU)**GPU/Geforce RTX3080
• DeepStream Version6.3
• NVIDIA GPU Driver Version (valid for GPU only) 530.41.03
I am trying to increase the number of file sources for DeepStream 3D Action Recognition App.
A few places were updated.

Added two file urls in
deepstream_triton_action_recognition_config.txt

[action-recognition]
# stream/file source list
uri-list=file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_walk.mov;file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_ride_bike.mov;file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_run.mov;file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_push.mov;file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_720p.mp4;file:///opt/nvidia/deepstream/deepstream/samples/streams/sample_1080p_h265.mp4

Change model engine file name and change batch-size in config_infer_primary_3d_action.txt

model-engine-file=./resnet18_3d_rgb_hmdb5_32.etlt_b6_gpu0_fp16.engine
batch-size=6

Change network-input-shape in config_preprocess_3d_custom.txt
network-input-shape= 6;3;32;224;224

Then I like to increase display window size and changed to
tiler-height=1080
tiler-width=1920

But still the same.

How to increase number of sources and display size?

  1. do you mean the output video only has 4 sub windows?
  2. how do you know the output window is not 1920x1080? please dump the media pipeline to check by this method.

Yes only 4 sub windows. I changed to 6 sources. But still 4. The size is not changed.

please refer to the following steps:

  1. set batch-size to 6 in config_infer_primary_3d_action.txt.
  2. run “./deepstream-3d-action-recognition -c deepstream_action_recognition_config.txt” to generate to batch-size 6 engine.
  3. set max_batch_size to 6 in config.pbtxt, and set max_batch_size to 6 in config_triton_infer_primary_3d_action.txt
  4. run ./deepstream-3d-action-recognition -c
    config_triton_infer_primary_3d_action.txt

Yes it works.
I changed network-input-shape to 6;3;32;224;224. Actually not necessary.
So fast, can achieve very high frame rate.

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