Automating Nutritional Claim Verification: The Role of OCR and Machine Learning in Enhancing Food Label Transparency

Consumer demand for health and trust in food products has widely gained attention, leading many researchers to call for rich information about food label nutritional content to be transparent and accurate. But nutritional claims such as “organic” or “sugar-free” still have verification problems because misleading labels may counter the trustworthiness and safety of a consumer. In this paper, we propose a data-driven approach to model verification of nutritional claims automatically through Optical Character Recognition and classification algorithms. Furthermore, EasyOCR, a deep-learning OCR tool, is used to analyze the ingredient data from the food labels against the nutritional standards defined by the USDA. Classification accuracy: through suggested models in machine learning such as SVM and Decision Trees, they classify the binary and multi-category claims. The SVM model achieved high strength with binary verification of claim accuracy and 90% accuracy, while Decision trees explained complex, multi-featured claims. The findings also revealed that this strategy will achieve providing consumers with improved accessibility to reliable nutritional information, thereby bolstering consumer trust in food labeling and smart eating. Not only does this approach streamline the verification process, but it also represents a significant advancement in food safety and transparency, as well as an important initial step towards the automation of claim verification technologies.