一、车辆识别
1、安装百度ai的python SDK
使用到python调用百度ai的接口，官网下载相对应的SDK：https://ai.baidu.com/sdk#vis
解压，python setup.py install
2、调用接口：
调用代码，其中xxxxx这3项需要使用百度ai后台提供相应内容，目前免费开放，接口每天有使用限制，直接注册创建一个应用即可。
每种识别场景都有各自对应的调用语句和返回值，我用的是“车辆识别”
https://ai.baidu.com/docs#/ImageClassify-Python-SDK/top
https://ai.baidu.com/docs#/ImageClassify-Python-SDK/496d7920
[cc lang=”python”]
from aip import AipImageClassify
import json
APP_ID = ‘xxxxx’
API_KEY = ‘xxxxx’
SECRET_KEY = ‘xxxxx’
client = AipImageClassify(APP_ID, API_KEY, SECRET_KEY)
def get_file_content(filePath):
with open(filePath, ‘rb’) as fp:
return fp.read()image = get_file_content(‘mycar.jpg’)
dict=client.carDetect(image)
print json.dumps(dict, ensure_ascii=False)
[/cc]
找一张汽车的图片，命名为mycar.jpg，例如下图：
百度ai接口返回的json如下，结果还算是准的，根据置信度有多个选择：
| { “log_id”: 7957146212839390462, “location_result”: { “width”: 823, “top”: 105, “left”: 112, “height”: 379 }, “result”: [{ “score”: 0.99940371513367, “name”: “本田雅阁”, “year”: “2013-2017” }, { “score”: 1.8783097402775e-05, “name”: “本田飞度”, “year”: “2004-2016” }, { “score”: 1.5786132280482e-05, “name”: “本田歌诗图”, “year”: “2010-2016” }, { “score”: 1.4914214261808e-05, “name”: “本田凌派”, “year”: “2013-2017” }, { “score”: 1.0179050150327e-05, “name”: “本田思铂睿”, “year”: “2009-2017” }], “color_result”: “黑色” } |
3、“车辆识别”相关请求和反馈参数如下：
车辆识别 请求参数详情
| 参数名称 | 是否必选 | 类型 | 默认值 | 说明 |
|---|---|---|---|---|
| image | 是 | string | 图像数据，base64编码，要求base64编码后大小不超过4M，最短边至少15px，最长边最大4096px,支持jpg/png/bmp格式 | |
| top_num | 否 | string | 返回预测得分top结果数，默认为5 | |
| baike_num | 否 | string | 0 | 返回百科信息的结果数，默认不返回 |
车辆识别 返回数据参数详情
| 字段 | 是否必选 | 类型 | 说明 |
|---|---|---|---|
| log_id | 否 | uint64 | 唯一的log id，用于问题定位 |
| color_result | 是 | string | 颜色 |
| result | 否 | car-result() | 车型识别结果数组 |
| +name | 否 | string | 车型名称，示例：宝马x6 |
| +score | 否 | double | 置信度，示例：0.5321 |
| +year | 否 | string | 年份 |
| +baike_info | object | 否 | 对应识别结果的百科词条名称 |
| ++baike_url | string | 否 | 对应识别结果百度百科页面链接 |
| ++image_url | string | 否 | 对应识别结果百科图片链接 |
| ++description | string | 否 | 对应识别结果百科内容描述 |
| location_result | 否 | string | 车在图片中的位置信息 |
二、身份证识别
http://ai.baidu.com/docs#/OCR-API/7e4792c7
使用bash方式，通过api key、secret key获取access_token，用于调用百度api
curl -i -k ‘https://aip.baidubce.com/oauth/2.0/token?grant_type=client_credentials&client_id=XXXXX&client_secret=XXXXX‘
获取token信息如下，留意access_token那行，下一步会用到：
{
“refresh_token”: “”,
“expires_in”: 2592000,
“session_key”: “”,
“access_token”: “XXXXX”,
“scope”: “public vis-ocr_ocr vis-classify_dishes brain_ocr_scope vis-classify_car brain_ocr_general brain_ocr_general_basic vis-ocr_business_license brain_ocr_webimage brain_all_scope brain_ocr_idcard brain_ocr_driving_license brain_ocr_vehicle_license vis-ocr_plate_number vis-classify_animal vis-classify_plant brain_solution brain_ocr_plate_number brain_ocr_accurate brain_ocr_accurate_basic brain_ocr_receipt brain_ocr_business_license brain_object_detect brain_realtime_logo brain_dish_detect brain_car_detect brain_animal_classify brain_plant_classify brain_solution_iocr brain_ingredient brain_ocr_handwriting brain_ocr_passport brain_ocr_vat_invoice brain_advanced_general_classify brain_custom_dish brain_numbers brain_ocr_train_ticket brain_ocr_taxi_receipt brain_poi_recognize vis-ocr_\u8f66\u8f86vin\u7801\u8bc6\u522b vis-ocr_\u5b9a\u989d\u53d1\u7968\u8bc6\u522b brain_ocr_vin brain_ocr_quota_invoice wise_adapt lebo_resource_base lightservice_public hetu_basic lightcms_map_poi kaidian_kaidian ApsMisTest_Test\u6743\u9650 vis-classify_flower lpq_\u5f00\u653e cop_helloScope ApsMis_fangdi_permission smartapp_snsapi_base iop_autocar oauth_tp_app smartapp_smart_game_openapi oauth_sessionkey smartapp_swanid_verify smartapp_opensource_openapi smartapp_opensource_recapi”,
“session_secret”: “”
}
找一张身份证图片，命名为sfz1.jpg，读取身份证图片，填入上一步生成的access token，调用百度api：
[cc lang=”python”]
# coding:utf-8
import base64
import urllib, urllib2, sysaccess_token = ‘XXXXX‘
url = ‘https://aip.baidubce.com/rest/2.0/ocr/v1/idcard?access_token=’ + access_token
f = open(r’sfz1.jpg’, ‘rb’)
img = base64.b64encode(f.read())
params = {“image”: img, “id_card_side”: “front”}
params = urllib.urlencode(params)
request = urllib2.Request(url, params)
request.add_header(‘Content-Type’, ‘application/x-www-form-urlencoded’)
response = urllib2.urlopen(request)
content = response.read()
if (content):
print(content)
[/cc]
返回的json信息如下：
| { “log_id”: 112233, “words_result_num”: 6, “direction”: 0, “image_status”: “normal”, “words_result”: { “住址”: { “location”: { “width”: 472, “top”: 516, “height”: 101, “left”: 292 }, “words”: “广东省广州市广州塔顶楼” }, “出生”: { “location”: { “width”: 314, “top”: 421, “height”: 40, “left”: 291 }, “words”: “19990909” }, “姓名”: { “location”: { “width”: 127, “top”: 235, “height”: 50, “left”: 292 }, “words”: “张三” }, “公民身份号码”: { “location”: { “width”: 613, “top”: 744, “height”: 42, “left”: 475 }, “words”: “123456789000000000” }, “性别”: { “location”: { “width”: 36, “top”: 333, “height”: 41, “left”: 290 }, “words”: “男” }, “民族”: { “location”: { “width”: 32, “top”: 336, “height”: 37, “left”: 520 }, “words”: “汉” } } } |