This repository contains the code for Building a simple Keras + deep learning REST API, published on the Keras.io blog with minor changes.

The method covered here is intended to be instructional. It is not meant to be production-level and capable of scaling under heavy load.

I assume you already have Keras (and a supported backend) and TensorFlow installed on your system. From there you need to install Flask and requests:

$ pip install flask gevent requests

*NOTES

• the original code has had all import statements of from keras updated to tensorFlow.keras
• ignore cudart64_110.dll GPU error
• may need to:

$ pip uninstall tf-nightly

and

$ pip install tensorflow --upgrade --force-reinstall

if receiving keras.utils.generic_utils module AttributeError.

Next, clone the repo:

$ git clone https://github.com/artoflearning/Image-Classification-Deep-CNN-Model-Keras-TensorFlow---Flask-.git

Below you can see the image we wish to classify, a dog, but more specifically a Shih-Tzu:

The Flask + Keras server can be started by running:

$ python run_keras_server.py 
Using TensorFlow backend.
 * Loading Keras model and Flask starting server...please wait until server has fully started
...
 * Running on http://127.0.0.1:5000

You can now access the REST API via http://127.0.0.1:5000.

Requests can be submitted via cURL:

$ curl -X POST -F [email protected] 'http://localhost:5000/predict'
{
  "predictions": [
    {
      "label": "Shih-Tzu", 
      "probability": 0.7272
    }, 
    {
      "label": "Pekinese", 
      "probability": 
  "success": true
}

Or programmatically:

$ python simple_request.py 
1. Shih-Tzu: 0.7272
2. Pekinese: 0.1508
3. Lhasa: 0.0415
4. Japanese_spaniel: 0.0225
5. Sussex_spaniel: 0.0165