tensorflow2 serving

tensorflow 模型训练部署为tfserving 服务

有以下三部
1 模型训练保存为savemodel
2 保存的模型在docker 部署服务。
3 在调用http 接口，进行模型推理。

1 模型训练保存为 models.save_model

import sys 
import tensorflow as tf
from tensorflow import keras
import numpy as np
import matplotlib.pyplot as plt
import os
import subprocess

#1 创建模型
(train_images, train_labels), (test_images, test_labels) = keras.datasets.fashion_mnist.load_data()

# scale the values to 0.0 to 1.0
train_images = train_images / 255.0
test_images = test_images / 255.0

# reshape for feeding into the model
train_images = train_images.reshape(train_images.shape[0], 28, 28, 1)
test_images = test_images.reshape(test_images.shape[0], 28, 28, 1)

class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
               'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']

print('\ntrain_images.shape: {}, of {}'.format(train_images.shape, train_images.dtype))
print('test_images.shape: {}, of {}'.format(test_images.shape, test_images.dtype))
# 2 训练和评估模型
model = keras.Sequential([
  keras.layers.Conv2D(input_shape=(28,28,1), filters=8, kernel_size=3, 
                      strides=2, activation='relu', name='Conv1'),
  keras.layers.Flatten(),
  keras.layers.Dense(10, name='Dense')
])
model.summary()

testing = False
epochs = 5

model.compile(optimizer='adam', 
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=[keras.metrics.SparseCategoricalAccuracy()])
model.fit(train_images, train_labels, epochs=epochs)

test_loss, test_acc = model.evaluate(test_images, test_labels)
print('\nTest accuracy: {}'.format(test_acc))

# 3 保存模型 为 SaveModel 格式
MODEL_DIR = './model'
version = 1
export_path = os.path.join(MODEL_DIR, str(version))
print('export_path = {}\n'.format(export_path))
tf.keras.models.save_model(
    model,
    export_path,
    overwrite=True,
    include_optimizer=True,
    save_format=None,
    signatures=None,
    options=None
)

print('\nSaved model:')

4检查模型
使用saved_model_cli命令查看savemodel保存后的的模型和方法

python D:\pythonapp\anacondas\envs\torchenv\Lib\site-packages\tensorflow\python\tools\saved_model_cli.py show --dir model\1 --all
MetaGraphDef with tag-set: ‘serve’ contains the following SignatureDefs:

signature_def['__saved_model_init_op']:
  The given SavedModel SignatureDef contains the following input(s):
  The given SavedModel SignatureDef contains the following output(s):
    outputs['__saved_model_init_op'] tensor_info:
        dtype: DT_INVALID
        shape: unknown_rank
        name: NoOp
  Method name is:

signature_def['serving_default']:
  The given SavedModel SignatureDef contains the following input(s):
    inputs['Conv1_input'] tensor_info:
        dtype: DT_FLOAT
        shape: (-1, 28, 28, 1)
        name: serving_default_Conv1_input:0
  The given SavedModel SignatureDef contains the following output(s):
    outputs['Dense'] tensor_info:
        dtype: DT_FLOAT
        shape: (-1, 10)
        name: StatefulPartitionedCall:0
  Method name is: tensorflow/serving/predict

2 docker 运行tfserving 服务

1 docker 拉取tfserving 镜像。
docker pull tensorflow/serving

2 将上面保存的模型放到某个目录下
我是在windows下训练的模型，将保存在model路径下的模型 放在了/opt/tfserving下。

3 . 构建模型和tserving 的链接，启动服务。

docker run -p 8501:8501   --mount type=bind,source=/opt/tfserving/model/,target=/models/model   -e MODEL_NAME=model -t tensorflow/serving

4 模型提供的服务请求默认为 http://localhost:8501/v1/models/model:predict
在windows 下请求该接口。结果如图

import requests
iheaders = {"content-type": "application/json"}
json_response = requests.post('http://192.168.10.100:8501/v1/models/model:predict', data=data, headers=headers)
predictions = json.loads(json_response.text)['predictions']

for i in range(0,3):
    show(i, 'The model thought this was a {} (class {}), and it was actually a {} (class {})'.format(
    class_names[np.argmax(predictions[i])], np.argmax(predictions[i]), class_names[test_labels[i]], test_labels[i]))

3 完整代码

1 模型训练保存为savemodel （可以使用saved_model_cli命令查看savemodel保存后的的模型和方法）后面预测时会使用到
2 保存的模型在docker 部署服务。（由于我的是window训练，linux服务，不能实现热更新，模型默认加载最新模型，也可以指定加载）。
3 在调用http 接口，进行模型推理。

完整代码

"""
part one 模型训练保存
"""
import sys 
import tensorflow as tf
from tensorflow import keras
import numpy as np
import matplotlib.pyplot as plt
import os
import subprocess

(train_images, train_labels), (test_images, test_labels) = keras.datasets.fashion_mnist.load_data()

# scale the values to 0.0 to 1.0
train_images = train_images / 255.0
test_images = test_images / 255.0

# reshape for feeding into the model
train_images = train_images.reshape(train_images.shape[0], 28, 28, 1)
test_images = test_images.reshape(test_images.shape[0], 28, 28, 1)

class_names = ['T-shirt/top', 'Trouser', 'Pullover', 'Dress', 'Coat',
               'Sandal', 'Shirt', 'Sneaker', 'Bag', 'Ankle boot']

print('\ntrain_images.shape: {}, of {}'.format(train_images.shape, train_images.dtype))
print('test_images.shape: {}, of {}'.format(test_images.shape, test_images.dtype))
model = keras.Sequential([
  keras.layers.Conv2D(input_shape=(28,28,1), filters=8, kernel_size=3, 
                      strides=2, activation='relu', name='Conv1'),
  keras.layers.Flatten(),
  keras.layers.Dense(10, name='Dense')
])
model.summary()

testing = False
epochs = 5

model.compile(optimizer='adam', 
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=[keras.metrics.SparseCategoricalAccuracy()])
model.fit(train_images, train_labels, epochs=epochs)

test_loss, test_acc = model.evaluate(test_images, test_labels)
print('\nTest accuracy: {}'.format(test_acc))

MODEL_DIR = './model'
version = 1
export_path = os.path.join(MODEL_DIR, str(version))
print('export_path = {}\n'.format(export_path))
tf.keras.models.save_model(
    model,
    export_path,
    overwrite=True,
    include_optimizer=True,
    save_format=None,
    signatures=None,
    options=None
)

print('\nSaved model:')


"""
part two
模型服务部署
	此部分在Linux完成
docker run -p 8501:8501   --mount type=bind,source=/opt/tfserving/model/,target=/models/model   -e MODEL_NAME=model -t tensorflow/serving
"""

"""
part three 模型推理测试
"""
def show(idx, title):
    plt.figure()
    plt.imshow(test_images[idx].reshape(28,28))
    plt.axis('off')
    plt.title('\n\n{}'.format(title), fontdict={'size': 16})

import json
data = json.dumps({"signature_name": "serving_default", "instances": test_images[0:1].tolist()})
print('Data: {} ... {}'.format(data[:50], data[len(data)-52:]))

import requests
headers = {"content-type": "application/json"}
json_response = requests.post(r'http://192.168.10.100:8501/v1/models/model:predict', data=data, headers=headers)
predictions = json.loads(json_response.text)['predictions']
show(0, 'The model thought this was a {} (class {}), and it was actually a {} (class {})'.format(
  class_names[np.argmax(predictions[0])], np.argmax(predictions[0]), class_names[test_labels[0]], test_labels[0]))