JDhyeok / flask-ml-model-api

Deploy deep learning model using Flask

falsk-ml-model-api

Deploy sentiment anaysis deep learning model using flask. It allows you using the result of model prediction through REST API

Installation

Source Code

$ git clone https://github.com/JDhyeok/falsk-ml-model-api.git

Project Setup/Start

Dependency

• Docker (Optional)
• Tensorflow
• Keras
• Python >= 3.x

Option1 - Local

• Setup virtual environment

  $ virtualenv venv

• Start python virtual environment

  $ .\venv\Script\activate
  # or source .\venv\Script\activate

• Install dependencies

  $ pip install -r requirements

• Start Flask server

  $ cd app
  $ flask run

Option2 - Docker

• Build docker image

  $ docker-compose build

• Start docker container

  $ docker-compose up

Request

You can use cURL examples below or use third party application such as PostMan.

• api test

  $ curl http://localhost:5000/api

• Predict sentiment score

  $ curl -X GET -H "Content-Type: application/json" --data '{"contents":"It is the best movie i've ever seen in my whole life"}' http://localhost:5000/api/predict

• Response JSON example

  {
      sentiment: "Positive"
      score: 95.1245
  }

Ref.

http://ai.stanford.edu/~amaas/papers/wvSent_acl2011.pdf

https://www.kaggle.com/lakshmi25npathi/sentiment-analysis-of-imdb-movie-reviews