-
Flask website
Model and Code
- Model and slides are here
- Dataset and Code can be found on GitHub: https://github.com/czlucius/bbcs-scanyourfood
- Website: http://168.138.177.217:28508/
Inspiration
One of our teammates, Lucas, has a friend with many allergies, and has to avoid a variety of foods.
What it does
It scans the image fed into it, processes it, and returns the most likely category. It classifies the image into multiple categories.
How we built it
We used TensorFlow and Keras for training the model, and predicting the categories. We used Flask to build a very simple web app for displaying the results. The server is hosted on an Oracle Cloud Instance running Ubuntu Linux.
Challenges we ran into
I couldn't get GPU acceleration to work at first, and spent a really long time to set up, installing multiple libraries(I did get it working with DirectML though) At one point, validation accuracy was too low. Seth, Zayan, and Jed helped me
Accomplishments that we're proud of
Getting the classifier to work, increasing accuracy through techniques like regularisation
What we learned
Deep/machine learning in general, the techniques used, TensorFlow, Keras and more
What's next for ScanYourFood
Scanning for more categories
Built With
- cnn
- keras
- python
- tensorflow
Log in or sign up for Devpost to join the conversation.