Help Needed: Running PeopleNet Model Live on Webcam with BBox Output in DeepStream

Accelerated Computing Intelligent Video Analytics DeepStream SDK
,
1

Hello,

I am working on a project where I need to run the PeopleNet model live on my webcam on my Jetson AGX ORIN 64gb dev using the NVIDIA DeepStream SDK. My goal is to obtain a real-time output that includes bounding box data for detected persons. However, I am facing challenges in setting up the environment and extracting the desired output. I would greatly appreciate any guidance or suggestions. Below are the details of my setup and the issues I am encountering:

  • Hardware Platform: [Jetson AGX ORIN 64gb dev]

  • DeepStream Version: [6.4.0-1]

  • JetPack Version: [6.0, 36.2.0-20231130105725]

  • TensorRT Version: [8.6.2.3]

  • I have downloaded all NVIDIA pretrained models and their configuration files. The models are located in /opt/nvidia/deepstream/deepstream-6.4/samples/configs/tao_pretrained_models/deepstream_reference_apps/deepstream_app_tao_configs.

  • Before running the PeopleNet model, I configured my camera in deepstream_app_source1_peoplenet.txt. Under [source0], I set the path to "uri=v4l2:///dev/video0" and adjusted the resolution (width and height) and FPS to match my webcam’s specifications.

  • To edit the deepstream_app_source1_peoplenet.txt file, I used the command sudo gedit /opt/nvidia/deepstream/deepstream-6.4/samples/configs/tao_pretrained_models/deepstream_reference_apps/deepstream_app_tao_configs/deepstream_app_source1_peoplenet.txt, as editing requires sudo privileges.

  • To launch the PeopleNet detection model, I navigate to the directory with cd "/opt/nvidia/deepstream/deepstream-6.4/samples/configs/tao_pretrained_models/deepstream_reference_apps/deepstream_app_tao_configs" and then run the command sudo deepstream-app -c deepstream_app_source1_peoplenet.txt.

Requirement details:

  • My goal is to run the PeopleNet model live on my webcam using the DeepStream SDK.
  • I am looking to extract and output bounding box data for detected objects in real-time. However, I am encountering difficulties in accessing and displaying the bounding box data.
  • I would greatly appreciate any specific instructions or code examples that could assist in achieving this, particularly in the context of using the standard DeepStream SDK (not the Python version) and the configurations I have made.

Thank you in advance for any help or pointers you can provide.

3

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

Since you are using a webcam, you can refer to samples\configs\deepstream-app\source1_usb_dec_infer_resnet_int8.txt to learn how to config the source parameter.

4

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