Skip to content

TensorFlow api for image recognition wrapped in a docker container and exposed as SaaS

Notifications You must be signed in to change notification settings

DarioBernardo/Image_recognition_as_a_service

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Image recognition as a service

This project uses the Google Tensorflow api to classify any picture on the internet

It uses links as the following:

http://www.hvac2000.com/wp-content/uploads/2016/05/Orange-Images-6.jpg

and return a JSON classifying the image, like the following:

{ "orange": 0.9540771842002869, "lemon": 0.0071089621633291245, "banana": 0.002119163516908884, "strawberry": 0.0005894483765587211, "pineapple, ananas": 0.0005539731937460601 }

To launch the microservice, just enter the folder and type

docker-compose up

Before interacting with the app, wait for the following message:

python_app_1 | Succesfully downloaded inception-2015-12-05.tgz 88931400 bytes.

The app needs to download the deep network model from the internet first. It will take a couple of seconds.

To get an image classification, you can use the /classify entrypoing with the "link" attribute to specify an image URL. Like the example below.

http://localhost/classify?link=http%3A%2F%2Fwww.hvac2000.com%2Fwp-content%2Fuploads%2F2016%2F05%2FOrange-Images-6.jpg

Alternatively, you can just go to http://localhost and use the GUI there on that page.

NOTE: The software do not work with HTTPS links! Furthermore it works with 'jpg' images only.

MIT Lincence

About

TensorFlow api for image recognition wrapped in a docker container and exposed as SaaS

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published