图像分类

from openvino.inference_engine import IECore
import numpy as np
import cv2 as cv

ie = IECore()                                                   #推理引擎
for device in ie.available_devices:                             #可用设备
    print(device)

with open('imagenet_classes.txt') as f:
    labels = [line.strip() for line in f.readlines()]

model_xml = "resnet18.xml"
model_bin = "resnet18.bin"

net = ie.read_network(model=model_xml, weights= model_bin)     #读取模型
input_blob = next(iter(net.input_info))                        #输入
out_blob = next(iter(net.outputs))                             #输出

print("input layout: ",net.input_info[input_blob].layout)           #模型输入格式NCHW
print("input precision: ",net.input_info[input_blob].precision)     #模型精度F32
print("input shape: ",net.input_info[input_blob].tensor_desc.dims)  #模型输入shape[1,3,224,224]

print("output layout: ",net.outputs[out_blob].layout)         #模型输出格式NC
print("output precision: ",net.outputs[out_blob].precision)   #模型精度F32
print("output shape: ",net.outputs[out_blob].shape)           #模型输出shape[1,1000]

n, c, h, w = net.input_info[input_blob].input_data.shape      #模型输入shape
print(n, c, h, w)

src = cv.imread("001.jpg")                                    #HWC
image = cv.resize(src, (w, h))                                #改变大小到输入大小
image = np.float32(image) / 255.0                             #归一化
image[:, :, ] -= (np.float32(0.485), np.float32(0.456), np.float32(0.406))
image[:, :, ] /= (np.float32(0.229), np.float32(0.224), np.float32(0.225))
image = image.transpose(2, 0, 1)                              #HWC->CHW

exec_net = ie.load_network(network=net, device_name="CPU")    #加载模型
res = exec_net.infer(inputs={input_blob:[image]})             #推理

res = res[out_blob]                                           #推理结果 (1,1000)
print(res.shape)
label_index = np.argmax(res, 1)[0]                            #获取最大概率
print(label_index, labels[label_index])
cv.putText(src, labels[label_index], (50, 50), cv.FONT_HERSHEY_SIMPLEX, 1.0, (0, 0, 255), 2, 8)
cv.imshow("image classification", src)
cv.waitKey(0)