问题描述

在使用 onnxruntime 推理,深度学习模型时,报了个错:

onnxruntime.capi.onnxruntime_pybind11_state.InvalidArgument: [ONNXRuntimeError] : 2 : INVALID_ARGUMENT : Unexpected input data type. Actual: (N11onnxruntime17PrimitiveDataTypeIdEE) , expected: (N11onnxruntime17Pri
mitiveDataTypeIfEE)

代码如下:

import onnxruntime
import cv2
import numpy as np

import ipdb

def transfrom_img(img_path):
    img=cv2.imread(img_path)
    mean=np.mean(img,axis=(0,1))
    std=np.std(img,axis=(0,1))
    final=np.transpose((img-mean)/std,(2,0,1))
    final=np.expand_dims(final,axis=0)
    return final

ort_session=onnxruntime.InferenceSession('../kp2d.onnx')

img_path='/home/chiebotgpuhq/intel/openvino_2021.2.185/deployment_tools/my_test/1.jpg'

img=transfrom_img(img_path)
ort_inputs={ort_session.get_inputs()[0].name: img}
ort_outs=ort_session.run(None,ort_inputs)

ipdb.set_trace()

print(ort_outs)

解决方法

将输入的numpy array转为float32,即把上代码中final改为final.astype(np.float32)

参考

• https://www.jianshu.com/p/fa2ea3750554