python 版本tensorrt推理

可以使用 TensorRT Python API 进行推理，需要安装 pycuda 和 tensorrt 库。下面是一个简单的例子：

import pycuda.driver as cuda
import pycuda.autoinit
import tensorrt as trt

# Load the serialized engine from file
with open(engine_path, "rb") as f, trt.Runtime(TRT_LOGGER) as runtime:
    engine = runtime.deserialize_cuda_engine(f.read())

# Create context for inference
with engine.create_execution_context() as context:
    # Allocate device memory for inputs and outputs
    inputs, outputs, bindings = allocate_buffers(engine)
    
    # Set input data
    inputs[0].host = input_data
    
    # Do inference
    trt_outputs = do_inference(context, bindings=bindings, inputs=inputs, outputs=outputs)
    
    # Get output data
    output_data = trt_outputs[0].host