体验京东云 Serverless+AI 人脸属性识别

Alt

云原生计算基金会CNCF（Cloud Native Computing Foundation, CNCF）Serverless Whitepaper v1.0对无服务器计算作了如下定义：

Serverless computing refers to the concept of building and running applications that do not require server management. It describes a finer-grained deployment model where applications, bundled as one or more functions, are uploaded to a platform and then executed, scaled, and billed in response to the exact demand needed at the moment.

"RightScale 2018 State of the Cloud Report" 中指出， Serverless 的年增长率达到了 75%，已经远超过 Container-as-a service 的 36%，位列第一，是增长速度最快的扩展云服务。

顺应当下容器技术、IoT、AI、5G、区块链等新技术的快速发展，技术上对去中心化、轻量虚拟化、细粒度计算等需求愈发强烈，在此趋势之上，Serverless 必将快速发展并得到应用。

对于开发者而言，Serverless 极大简化了编写代码和应用部署的整个过程，无需关心 DevOps，无需关心代码运行服务器及环境，更无需关心负载均衡，中间件和存储服务可以随时调用。

Q

Serverless 的优势有哪些？

A

无需运维：用户不需要再关心服务器资源及复杂运行环境；
弹性伸缩：根据请求并发量实时自动扩缩资源；
按需付费：为实际消耗资源付费，无占用不收费，运行成本更低；

云上提供一整套serverless FaaS+BaaS服务，帮助用户快速构建应用与服务。

今天，我们会通过快速组合京东云 Serverless + AI，实现一个简单的 API 接口，进行人脸属性识别和简单数据分析。带领大家一起来感受一下Serverless的轻量虚拟化。

Alt

关键词

Serverless Service：函数服务+API 网关

人工智能产品：AI 人脸识别

Step1：创建函数

创建 runtime 为 python2.7 的函数，将依赖库 jdcloud_apim_sdk、ai_sdk（京东云AI）和入口文件 index.py 打包上传，index.py 代码如下：

#coding=utf-8

import json
import base64
import os

from jdcloud_apim_sdk.core.credential import Credential
from jdcloud_apim_sdk.core.config import Config
from jdcloud_apim_sdk.core.const import SCHEME_HTTPS, SCHEME_HTTP

from ai_sdk.apis.ai_face_detect_attr_request import *
from ai_sdk.client.ai_Face_Detect_Attr_client import *

'''
AI
'''

def handler(event,context):
    if not bool(event):
        result = {
            'statusCode': 200,
            'headers': {},
            'body': "",
        }

        return result

    # get request data
    print('AI Request')
    # get request data
    body = event['detail']['body']
    body = json.loads(body)

    if type(body) == unicode:
      body = json.loads(body)

    # request
    access_key = 'XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX'
    secret_key = 'XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX'
    credential = Credential(access_key, secret_key)
    config = Config('aiapi.jdcloud.com', scheme=SCHEME_HTTPS)
    client = AiFaceDetectAttrClient(credential, config)

    header = dict()
    parameters = {
            'return_landmark':0,
            'return_attr':1,
            'face_detect_num':0
            }
    data = body.get('data', "")
    imageBase64Data = 'imageBase64=' + data.decode("utf-8")
    s = imageBase64Data.encode('utf-8')

    ai_face_detect_attr_request = AiFaceDetectAttrRequest(parameters, s, header)
    AiFaceDetectAttr_response = client.send(ai_face_detect_attr_request)

    respBody = {'code': 0, 'data': AiFaceDetectAttr_response.json()}
    resp = json.dumps(respBody)
    result = {
        'statusCode': 200,
        'headers': {},
        'body': resp,
    }

    # statistics
    resp = AiFaceDetectAttr_response.json()
    if resp['result']['status'] == 0:
      attrs = json.loads(resp['result']['result'])
      if len(attrs['faces']) > 0:
        if attrs['faces'][0]['faceAttribute']['gender'] == 0:
          print("lookup ai attr success, this is a woman")
        else:
          print("lookup ai attr success, this is a man")

    return result