dragofireup / web-food-classifier

A web application which uses a Neural Network, trained on Food-101, to predict the class of an image

Food Classifier Web App

This is a web application which allows a user to upload an image and the get the probabilities for top 3 classes from the neural network.

The Live Web App can be found here

Details about the Neural Network

Dataset

• Food-101

Model

• Fine Tuned ResNet50

Training Environment

• Python: version 3.6
• fastai: version 1.0.61
• PyTorch: version 1.5.1
• CUDA: version 10.2

Deployment Environment

• Python: version 3.6
• fastai: version 1.0.61
• PyTorch: version 1.5.1 (cpu version)
• Flask: version 1.1.2
• Heroku

Want to run on your system??

Make sure you have either the training or deployment environment set up.

To setup the deployment environment

pip install -r requirements.txt

Run using flask commands

To run the flask application

flask run

This runs your flask application on http://127.0.0.1:5000

To run the application on port 80

flask run -p 80

To run the application on your network IP address

flask run -h 0.0.0.0

Run using python

python app.py

This command will run the application on your network IP address on port 80.

You can access the site using 127.0.0.1 too even when running on network IP