This is a repo with a Triton Server deployment template. As a model example YOLOv8 was chosen, converted to TensorRT. Tested on Nvidia 3060.
Get your yolov8 detection model weigts exported to engine with ultralytics docs
docker run --gpus=all --rm -d -p8000:8000 -p8001:8001 -p8002:8002 -v/model_repo_path:/models nvcr.io/nvidia/tritonserver:23.07-py3 tritonserver --model-repository=/models
model_repo_path
- full path to your model_repository
23.07-py3 - version of your triton inference server
name_to_label_mapping
to fit your labelspython -m src.main