Nutrition tracking from photos of the food

A nutrition tracking app that gets nutrition numbers from photos of the food is now available on Github.

Nutrition tracking tracking used to be a lot of work. With this new Snap & Eat app, it’s as easy as taking a picture of your food.

Snap and Eat app screenshot

This amazing app was built in 48 hours at a Deep Learning Hackathon. The Snap & Eat app allows nutrition tracking from photos of the food. It also recommends meals. The creators won 1st place!

Deep Learning Hackathon 48h – Cotidiano (First place Project =D )

We believe nutrition tracking should be as simple as taking a picture.

Snap & Eat is a web application that tracks the user’s food intake by pictures. We use state-of-the-art deep learning techniques to recognize dishes, making instant nutrition estimates from the user’s meals.

The app also suggests meals based on the user’s income, and is capable of showing places nearby that serve those dishes.

The system is implemented in Pytorch using fastai lib, relying on Jupyter Notebooks for prototyping purposes. For the web app, we use Flask and Node.js.

 

Snap & Eat’s Model

We use an Aggregated Residual Convolutional Neural Network – ResNeXt-101 with 101 layers, pretrained on ImageNetdataset. We finetune the model on Food-101 dataset, with more than 100 thousand images of 101 types of dishes. We achieve a significant improvement on accuracy (71% in our work compared to 50.1% in Bossard et al., 2014).

test

For recomending new dishes, we use minimum ditance in an n-dimensional space of nutritional information that describe each dish.

Installation and usage

AWS AMI (Amazon Machine Images)

We provide a ready-to-go AMI of a p2 instance on AWS with all libraries and dependencies installed.

AMI ID ami-99442ee3
AMI Name snap-n-eat-ami

To launch an instance based on an AMI, see the AWS Reference. Don’t forget to git pull to get the the latest version of the code!

cd snapandeat
git pull origin master

To run all the services, the following four comands need to be run in the AWS instance, once in the snapandeat directory. We recommend using tmux to organize the terminals.

python src/services/food_predictor.py
python src/services/food_descriptor.py
python src/services/top_food.py
node src/webapp/server.js

To view and edit the Jupyter Notebooks, execute:

cd /home/ubuntu/snapandeat/src/notebooks
jupyter-notebook --no-browser --ip='*'

The default password for the notebooks is snapneat. Don’t forget to change it!

About the idea

According to the World Health Organization, worldwide obesity has nearly tripled since 1975. In the United States, almost 75% of the population is overweight and more than half of the population is obese (OECD). Today, many diseases that were preivously thought as hereditary are now shown to be seen conected to biological disfunction related to nutrition.

Although being healty and eating better is something the vast majority of the population want, doing so usually requires great effort and organization. The lack of an easy and simple way to track nutrition information about the food you eat can easily lead to low engagement. By providing a very easy and fun way to keep track of what the user eat, we can largely improve engagement, and directly atack on of the largest health problems in the world.