2018年11月3日 星期六

Call AutoML model found error expected one of ('authorized_user', 'service_account').



You can now run predictions on images using your custom vision model. 

python predict.py YOUR_LOCAL_IMAGE_FILE aiyvision-xxx ICNxxxxxxxxxxxxx

When run the command above before, you have to create a JSON file which have enough privileges to invoke the API for inferencing. 


https://console.cloud.google.com/iam-admin/serviceaccounts


Then Create another environment variable that points to the downloaded JSON key.
export GOOGLE_APPLICATION_CREDENTIALS=key-file //Linux set GOOGLE_APPLICATION_CREDENTIALS=key-file //Windows
 If see the error as following, you will need to do the 2 steps.

google.auth.exceptions.DefaultCredentialsError: The file D:\google_images_download\key.json. does not have a valid type. Type is None, expected one of ('authorized_user', 'service_account').



Reference site
https://cloud.google.com/automl/
https://cloud.google.com/iam/docs/creating-managing-service-accounts
https://thenewstack.io/get-started-with-google-cloud-automl-vision-for-image-classification/

2018年10月3日 星期三

加停小幫手/ParkingandRefuelin Privacy Policy

Parking and Refuelin
Use to find the refue station which around you and support parking space information as  台灣聯通, 中興嘟嘟房, 便利停車場, 福慧網, ViVipark

Permission user description

CAMERA: Required to be able to access the camera device. Use to take a photo for recognition.
READ_EXTERNAL_STORAGE: Allows an application to read from external storage. Use to pick a photo from your gallery.
INTERNET: Allows applications to open network sockets. Use to upload the photo to service for recognition via machine learning. 
 
Data collection
The App DO NOT collect any users data.
 
Link
All the links in the app that are secure and no any malicious programs. 
The link in app is used to link tutorial video on YouTube.
 
Contact us
If any privacy policy issue, please contact us by the mail address:
kof0018@gmail.com

2018年5月17日 星期四

Google Cloud vision API 實作Android App 猜吧 - 人工智慧猜圖神器

這是一個利用Google Cloud Vision API應用實例

參考網址 Google Cloud Platform — Cloud Vision API


Vision API主要有幾個分類:

LABEL:
使用這個API, 它會把圖中的物件或元素找出來,並給於一個值,這個值代表信心指數, 例如圖片中含有一台鋼琴,和杯子 label 在這張圖片裡出現的信心程度。假設拍辦公室場景,分析大概就會出現電燈,桌子,椅子,杯子,人...等等
 在App裡的setting裡的動物或交通工具就是用LABEL的API

Web:   
此API 會結合搜尋引擎功能,有點以圖搜圖的概念,App裡的All Detection,就是用此API,
例如,拍名人,拍車子,就會用這張圖來搜尋出這個名人叫什麼,車子是什麼型號

Landmark:
顧名思義, 用來拍知名地標,或知名人造建築,在App裡地標搜尋就是用這個API,
App裡個小功能,在判斷後結果,就會導到地點,適合旅遊時迷路拍一下XD

Logo:
用來搜尋產品logo,如LV, Gucci,可口可樂, 零食品牌等等
 
如果大大們用這個App在實際生活上有什麼酷炫的應用,也歡迎分享給小弟我,也可以加強或實作這個功能給更多人享用


Android App download link:
點擊下載   猜吧 Android App

2018年5月5日 星期六

How to get google-services.json from firebase

1)enter Firebase console    https://console.firebase.google.com/

2)Choose your project
 

3)Choose project setting



4)Download the json file



5)Put the json file under the app\ folder of your project