Publication

Abstract : The growing number of diet-related illnesses requires easily accessible precise nutritional information to be available. This study develops a sophisticated OCR solution with deep learning capabilities meant to assess packaged food contents through nutritional analysis. EasyOCR and PyTesseract extract key nutritional details consisting of calories, fats, sugars and sodium from both low-quality and distorted packaging labels. The model process collected information through a classification model containing the NOVA system together with a recently created Harmfulness Score (H Score) framework. The H Score represents a value that derives from weighted nutritional measurements which achieve validation from the conventional Nutri-Score system. The classification model enables precise measurement within one point for accurate data collection according to nutritional standards. The two-step assessment method improves food evaluation reliability so consumers and health professionals together with policy makers can base their dietary choices on factual information. Real-time analysis and upgraded label processing systems help users conduct better assessments on different packaging types. The developed research system helps users make healthier food choices by offering automated nutritional analysis capabilities for sustainable dietary practice implementation.