Overview / Usage

The main objective of our project was to create an application that would help facilitate a user with his or her diet. The application would do this by asking the user questions about his or her age, height, current weight, activity level and goal weight and use this information to set a specific amount of daily consumable calories for the user depending on if he/she wants to lose weight, gain weight or maintain the same weight. The user would keep track of these calories by using the application’s food recognition software which allows a user to scan a food item that he or she is about to eat and then recognizes this food item along with the number of calories of a typical serving size of that food item. These calories would be subtracted from the user’s daily consumable calories and the user’s remaining calories would be displayed.

Methodology / Approach

We trained a convolution neural network to make this project function. We used Google's Inception v3 model as the architecture to mimic because of it's large popularity and because of it's amazing results during the ImageNet Large Scale Visual Recognition Competition. We continued to improve and train our model with more specific data sets until we had above 80% accuracy.

Technologies Used

Intel Optimized TensorFlow, Android Camera2, Android Studio, Python, Java

Repository

https://github.com/PallabPaul/AI_Calorie_Counter_Demo

Other links

• Article