Description: PeopleNet is a deep learning model designed for real-time people detection and tracking, optimized for use with NVIDIA's DeepStream SDK 7.0. This model processes video streams to identify and analyze human presence in various environments, making it suitable for applications like smart surveillance and crowd management.

Author: Basil Shaji

Last Updated: Oct-14-2024

Organization: Karunya Institute of Technology and Sciences

Reference Docs:

1. https://docs.nvidia.com/metropolis/deepstream/6.0/dev-guide/text/DS_docker_containers.html
2. https://docs.nvidia.com/metropolis/deepstream/6.0/dev-guide/text/DS_ref_app_deepstream.html
3. https://docs.nvidia.com/metropolis/deepstream/6.0/dev-guide/text/DS_sample_custom_gstream.html
4. https://learn.nvidia.com/en-us/training/self-paced-courses

Have a quick reading through the Documentation.

1. Pull the docker container:

$ docker pull [nvcr.io/nvidia/deepstream:7.0-gc-triton-devel](<http://nvcr.io/nvidia/deepstream:7.0-gc-triton-devel>)

1. Connect the webcam and follow the below commands :)

#To run the docker as root:
docker run -it --rm --net=host --gpus all -e DISPLAY=$DISPLAY \\
--device /dev/video0 --device /dev/snd \\
-v /tmp/.X11-unix/:/tmp/.X11-unix \\
-v /etc/nvidia:/etc/nvidia \\
-v /opt/nvidia/deepstream/deepstream-7.0/samples/configs:/config \\
-v /opt/nvidia/deepstream/deepstream-7.0/samples/models:/models \\
-w /opt/nvidia/deepstream/deepstream-7.0/samples/configs/deepstream-app \\
[nvcr.io/nvidia/deepstream:7.0-gc-triton-devel](<http://nvcr.io/nvidia/deepstream:7.0-gc-triton-devel>)

1. Open another separate terminal and type these commands,

#To give access to the camera:
xhost +
gst-launch-1.0 v4l2src device=/dev/video0 ! videoconvert ! autovideosink

1. Inside the docker terminal, type this command to edit the deepstream model config file (source1_usb_dec_infer_resnet_int8.txt):

nano source1_usb_dec_infer_resnet_int8.txt

1. Changes made to the config file: