Health and Fitness Tracker Using Image Based Food Recognition and BMI Driven Recommendations

Because obesity and sedentary lifestyles have emerged as global issues, it calls for the advent of personalized health monitoring solutions. This paper attempts to build an end-to-end Health and Fitness Tracker, where food is being analyzed in real time, and health insights are given from the perspective of the user. In particular, the live detection of food is achieved using the YOLOv8 object detection model, which is capable of detecting food from either images captured in webcam mode or images uploaded directly from the user device. The system then cross-references the food-calorie dataset to arrive at the calculated caloric value of the item detected once the item has been detected. Meanwhile, the user can enter height and weight parameters for Body Mass Index (BMI) calculation, thus providing the application with a means of categorizing users under standard health categories: Underweight, Normal weight, Overweight, or Obese. The application suggests exercise routines and daily-level caloric intake limits according to their BMI classification. Using Tkinter for application development, the interface provides seamless access for navigation to food image analysis, calorie tracking in near real-time, BMI calculation, and personalized recommendations. The tracked and summed daily consumption logs allow users to work toward making better decisions about food intake and physical activity. Computer vision, data processing, and health analysis together provide a working definition of holistic digital fitness tracking. A ready, extensible system for those who would like to monitor, develop, and improve their eating and overall health using machine learning and human-focused design.