1

我已经使用 onnx 格式在 tensorrt 中实现了我的 Pix2Pix GAN 模型。但我不知道如何对 tensorRT 模型进行推理，因为 (3, 512, 512 ) 图像中模型的输入和输出也是 (3, 512, 512) 图像。样本没有清楚地显示如何从 tensorRT 引擎输入和输出图像。

这是我的代码：

   def get_image():
    img = _load_and_resize("/home/abdullah/Desktop/Profile.png")
    img = _shuffle_and_normalize(img)
    return img

def _load_and_resize(input_image_path):


    image_raw = Image.open(input_image_path)
    # convention (width, height) in PIL:

    image_resized = image_raw.resize(
        (512, 512), resample=Image.BICUBIC)
    image_resized = np.array(image_resized, dtype=np.float32, order='C')
    return image_resized


def _shuffle_and_normalize(image):

    image /= 255.0
    # HWC to CHW format:
    image = np.transpose(image, [2, 0, 1])
    # CHW to NCHW format
    image = np.expand_dims(image, axis=0)
    # Convert the image to row-major order, also known as "C order":
    image = np.array(image, dtype=np.float32, order='C')
    return image
def _reshape_output(output):

    return np.reshape(output, (512, 512, 3))
def main():
    """Create inference for generator for eye."""

    onnx_file_path = 'generator.onnx'
    engine_file_path = "generator.trt"

    output_shapes = [(1, 3,512, 512)]
    image =  get_image()
    print('IMage shape is ', image.shape)
    # Do inference with TensorRT

    with get_engine(onnx_file_path, engine_file_path) as engine, engine.create_execution_context() as context:

        inputs, outputs, bindings, stream = common.allocate_buffers(engine)
        print('Running inference on image {}...'.format(1))
        inputs[0].host = image
        print(inputs)
        trt_outputs = common.do_inference(context, bindings=bindings, inputs=inputs, outputs=outputs, stream=stream)

    # Before doing post-processing, we need to reshape the outputs as the common.do_inference will give us flat arrays.
    trt_outputs = [output.reshape(shape) for output, shape in zip(trt_outputs, output_shapes)]

    image = _reshape_output(trt_outputs)
    image = np.array(image, dtype = 'float32')
    plt.imshow(image)
    plt.show()
    print('Output shape is ', trt_outputs.shape)

但是输出并不像预期的那样。请告诉我，将图像输入到 tensorRT 引擎的最佳方式是什么，以及如何从中获得正确的输出。谢谢

4

0 回答 0