ccd97 / image-classify-server

Image classification using Tensorflow (Inception v3)

ImageClassification-Server

Image classification with Web API and UI.
It's is written in Django along with Tensorflow uing Google's Inception v3 model
The web interface is made using materializecss and jQuery
It is extension of this project.

Usage

To run the server on localhost:

$ pip3 install -r requirements.txt
$ python3 manage.py collectstatic
$ python3 manage.py runserver

Web Interface

The Web Interface can be found at http://tf-classify.herokuapp.com

Web API

You can classify using web API by sending a POST request at http://tf-classify.herokuapp.com/classify_image/classify/api/

Input

Parameter	Type	Description
image	file	Image file that you want to classify.
image64	text	Image in base64 form that you want to classify. Currently supports JPEG images only
k	text (optional, default=10)	Return top-k categories of the results. Must me string in integer format.

Note: you need to send either 'image' or 'image64'

Result

Parameter	Type	Description
success	bool	Whether classification was sucessfuly or not
confidence	category, float	pair of category and it's confidence

Note: category is not paramater name but string of the category.
Example: {"success": true, "confidence": { "mongoose": 0.87896, "hare": 0.00123 }}

Using Retrained Inception Model

• Retrain the model using your images. Refer here.
• Fork this repo
• Replace the generated graph and label files in /classify_image/inception_model/
• Deploy the Django project

Contribute

I am just beginner. If you find any bugs or want to improve the project, fell free to do it uisng pull request.