NutriSenseAI: An Intelligent System for Automated Nutrition Tracking

Diet intake monitoring is an important aspect in the management of healthy life that is unaffected of obesity, diabetes, and other diet related conditions. Most applications and websites require manual meal entry by users, which can be time-consuming and prone to errors, leading eventually to user despondency due to interest attrition over time. This paper proposes an AI-based system responsible for tracking food as a possible solution toward overcoming such challenges. The proposed system applies advanced artificial intelligence capable of automatically and accurately recognizing what people are eating from different multimodal inputs that include images and textual descriptions. We hereby used Google’s Gemini pro model for real-time recognizing the food item as being integrated with USDA data for nutritious content validation is done as much accurate as possible. We implemented a JSON-based approach which ensures standard data handling and effective integration with other systems. Through our app user will be able to resize portion, set custom diets, keep an eye on its diet in real- time environment. It supports voice input, making it easier and more fun to record food consumption. The effectiveness of the system was evaluated by using 50 different food images. The model reaches a mean absolute error of 9.8% in the calorie estimate and an accuracy of 92% in food recognition beating quite a few existing methods both in terms of accuracy as well as ease of use.