Foodie is a web app for detection of food/beverages from an image. Thousands of people post their food pics on social media daily. Automatic identification of food types from the pictures can be very useful on many occasions. A convolutional neural network is trained over Resnet on a dataset containing 11 categories of food viz. "Bread", "Dairy product", "Dessert", "Egg", "Fried food", "Meat", "Noodles/Pasta", "Rice", "Seafood", "Soup", "Vegetable/Fruit". A web application environment is setup using flask micro web framework to provide a simple GUI.

Clone the model. Download pretrained weights for Resnet on imagenet dataset. Train the model:

$ python3 train.py

Running the the web app

$ export FLASK_APP=predict_app.py
$ flask run --host=0.0.0.0

Open http://0.0.0.0:5000/static/predict.html from browser to see the application running on social machine.