Although food packaging comes with calorie labels, it is not very convenient for people to refer. Therefore, we propose a DEEP LEARNING based calorie tracking model to monitor people’s daily calorie intake. Based on an internally calculated parameter, our model also suggests the users, appropriate food, and relevant exercises. Along with this, we have a food-label tracker as well, which will suggest the user if a particular food item is fit for consumption or not.
DL-1 takes physical attributes such as Age, Gender, Height and Weight as the training features and gives user's health state as the output. In this model, the output is coded as: 0: Underweight, 1: Healthy, 2: Overweight, 3: Obese.
Dataset source: https://www.kaggle.com/freego1/bmi-data
DL-2 consists of two models, namely, food calorie model and activity calorie model. First model predicts the calorie intake value on consumption of given food, while the latter predicts calories burnt amount based on physical activity done (like walking, running, etc.).However, model was trained for limited data for prototype ,data can be expanded afterwards. Subtracting calorie intake and burnt data will give calorie consumed value.
Dataset_activity source: https://www.health.harvard.edu/diet-and-weight-loss/calories-burned-in-30-minutes-of-leisure-and-routine-activities
Dataset_food source: Kaggle, https://www.calories.info/food/, https://docs.google.com/spreadsheets/d/1snqE6leDkZlL61qQ4g-vUmiFjizJyN1OCVAhwWWKSm4/edit#gid=2024304766
DL-3 takes inputs from both DL-1 and DL-3 along with user's physical parameters. This model suggests healthy food and excercises like running and walking based on the calories intake.
DL-4 is a food label tracking model which takes a particular food item's nutritional information's image as its input. It then performs a binary classification on those images and lets the user know if that food item is healthy or unhealthy.
Dataset source: Dataset was created by taking images from several grocery websites like BigBasket, Flipkart, Amazon, etc.