/jZddlmZddlmZddlZddlmZddlmZGddeZ dS) ) annotations)PathN)check_requirements) BaseBackendc"eZdZdZd dZd d Zd S) TritonBackendzNVIDIA Triton Inference Server backend for remote model serving. Connects to and runs inference with models hosted on an NVIDIA Triton Inference Server instance via HTTP or gRPC protocols. The model is specified using a triton:// URL scheme. weight str | PathreturnNonectdddlm}|||_t |jdr!||jjdSdS)zConnect to a remote model on an NVIDIA Triton Inference Server. Args: weight (str | Path): Triton model URL (e.g., 'http://localhost:8000/model_name'). ztritonclient[all]r)TritonRemoteModelmetadataN)rultralytics.utils.tritonrmodelhasattrapply_metadatar)selfr rs c/home/longshao/multi-rider-rag/.venv/lib/python3.11/site-packages/ultralytics/nn/backends/triton.py load_modelzTritonBackend.load_modelsv .///>>>>>>&&v..  4:z * * 5    3 4 4 4 4 4 5 5im torch.Tensorlistct||S)aRun inference via the NVIDIA Triton Inference Server. Args: im (torch.Tensor): Input image tensor in BCHW format, normalized to [0, 1]. Returns: (list): Model predictions as a list of numpy arrays from the Triton server. )rcpunumpy)rrs rforwardzTritonBackend.forward$s(zz"&&((..**+++rN)r r r r )rrr r)__name__ __module__ __qualname____doc__rrrrr r sF 5 5 5 5 , , , , , ,rr ) __future__rpathlibrtorchultralytics.utils.checksrbaserr r$rrr*s#""""" 777777,,,,,K,,,,,r