A Deep Learning–Based Framework for Automated Food Detection, Nutrient Estimation, and Personalized Diet Monitoring

Growing awareness around healthy lifestyles and the demand for automated dietary monitoring have led to the development of intelligent food-recognition systems. Traditional calorie-tracking applications depend heavily on manual entry, which often introduces inaccuracies and reduces long-term user participation. To address these limitations, this work introduces NutriTrack, a unified framework that combines YOLO-based real-time food detection, internally mapped food classes, and Nutritionix API data to produce precise nutritional estimates. In addition, the system employs machine-learning techniques to deliver personalized meal recommendations tailored to individual user needs. By supporting multi-item detection, improving calorie estimation, and integrating a comprehensive tracking interface using Flask and SQLite, NutriTrack advances beyond conventional CNN-based approaches. Evaluation results indicate strong accuracy for single-item images and reliable performance in mixed-dish scenarios, offering an efficient and user-friendly solution for everyday nutritional assessment