We realized how much junk food is available at hackathons and how often you can lose track of how much you eat. We wanted to use IBM's visual recognition api to make tracking your foods as easy as a press of a button.
This application allows you to easily track your calorie consumption and send you text notifications when you reach calorie thresholds.
It takes images by using open-cv and sends them to IBM's visual-recognition api with custom train classifiers for typical hackathon food. The api can easily identify these foods and match them up to a value in calories. When you reach certain thresholds in calorie consumption you are sent a text via Twilio that tells you approximately how many calories you have consumed as well as a list of imgur links that you can visit to see what exactly added up to this value.
Training the classifiers for IBM's visual recognition was very tedious, but overall not that difficult. None of us had experience doing web development, so sending data to our website was a difficult task.
We're proud of the success rate of our custom classifiers for IBM's visual recognition api and of how streamlined our app is for the user.
We learned how to using an API reference to implement apis as well as trouble shooting issues by looking at source code for associated libraries.
If this project continues, it is important to make a simple gui so the application is more accessible to a wider audience of users. Additionally a mobile app should be made similar to Instagram so the app is accessible on the go.
opencv, bluemix, firebase, twilio, python, html5, css3, javascript, ibm-watson, imgur