NutriScan: A Mobile Application for Health-Conscious Food Choices

In response to the need for tools that streamline the extraction of nutritional information, this study presents a sophisticated technique to automatically detect nutrition tables and identify ingredients on packaged foods. Accurate tools are crucial for portion control and making educated food choices as awareness of healthy eating grows. We increase the accuracy of identifying nutrition tables and ingredients by improving the EfficientDet model. Regular expressions collect important nutritional information including sodium, carbs, fats, proteins, additives, and allergies, while PaddleOCR further refines text extraction. Customers can indicate dietary preferences and sensitivities (such as to sugar, fat, or salt) using a customized user preference system. Our technology assesses food items based on this, providing users with a Nutri-Score that is in line with their health objectives and personalized suggestions. The correctness and effectiveness of our method are confirmed by extensive testing.