0. if you have not already done so, generate an api token after logging in to https://org.ngc.nvidia.com/setup

1. login with nvidia token

$ docker login nvcr.io

Username: $oauthtoken Password: $YOUR_API_TOKEN 

2. pull docker container

$ docker pull nvcr.io/nvidia/deepstream:7.1-triton-multiarch

3. run docker container

$ docker run --gpus all -it --rm --network=host \ nvcr.io/nvidia/deepstream:7.1-triton-multiarch 

Note: after customizing the container, it will run as

docker run --gpus all -it --rm --network=host deepstream-7.1-custom 

4. list TensorRT engines

$ find /opt/nvidia/deepstream -name "*.engine"

5. edit config file

$ vi /opt/nvidia/deepstream/deepstream/samples/configs/deepstream-app/source1_file_dec_infer_resnet_int8.txt

a) set engine file:

model-engine-file=/opt/nvidia/deepstream/deepstream-7.1/samples/models/Primary_Detector/resnet18_trafficcamnet_pruned.onnx_b1_gpu0_int8.engine

b) set loop to true

c) add this section for profiling

[application] enable-perf-measurement=1 perf-measurement-interval-sec=5 

6. run pipeline

$ deepstream-app -c /opt/nvidia/deepstream/deepstream/samples/configs/deepstream-app/source1_file_dec_infer_resnet_int8.txt

6. optional : commit changes

docker ps docker commit $CONTAINER_ID deepstream-7.1-custom docker run --gpus all -it --rm --network=host deepstream-7.1-custom