一、安裝
必要:tensorflow,Keras
首次運行需要安裝:
1)下載模型權重 inception_v3_weights_tf_dim_ordering_tf_kernels.h5
路徑見前一篇
2)安裝h5py
pip install h5py
3)安裝PIL
遇到pip無法安裝,以pillow替代,見Stack Overflow

二、參數說明

分類結果:
ImageNet的1000種object,對應模型分類結果的1000 classes:
text: imagenet 1000 class id to human readable labels
https://github.com/cjyanyi/keras_deep_learning_tutorial/blob/master/imagenet1000_clsid_to_human.txt
三、代碼示例
import numpy as np
from keras.preprocessing import image
from keras.applications import inception_v3
img = image.load_img("xxx.jpg", target_size=(299, 299))
input_image = image.img_to_array(img)
input_image /= 255.
input_image -= 0.5
input_image *= 2.
# Add a 4th dimension for batch size (Keras)
input_image = np.expand_dims(input_image, axis=0)
# Run the image through the NN
predictions = model.predict(input_image)
# Convert the predictions into text
predicted_classes = inception_v3.decode_predictions(predictions, top=1)
imagenet_id, name, confidence = predicted_classes[0][0]
print("This is a {} with {:-4}% confidence!".format(name, confidence * 100))
input_image 是一個默認大小:1*299*299*3 的4維向量(列表)
